Robot Navigation System for Mapping of Non-GPS Accessible Terrain

Transcription

1 Robot Navigation System for Mapping of Non-GPS Accessible Terrain Project: Project Leader: Anja Soldo EE Team Members: Stephen Byrne ME Andrew Fishbaugh ME Ian Kunsch ME Michael Oswald ME Aneta Rozwadowska EE Vered Talmor EE Mentor/Customer: Dr. Crassidis Kate Gleason College of Engineering James E. Gleason Building Rochester Institute of Technology Rochester, NY Robot Navigation System for Mapping of...1

2 2 Non-GPS Accessible Terrain Recognize and Quantify Need Sponsor Background Project Mission Statement Robot Navigation Background MIMNS Scope Limitations Stakeholders Key Business Goals Top Level Critical Financial Parameters Financial Analysis Primary Market Secondary Market Teaser Topics Formal Statement of Work Concept Development Long List Group Drawings Short List FEASIBILITY ASSESSMENT (ELECTRICAL) Feasibility Assessment (Mechanical) Feasibility Strategy Feasibility Questions Initial Feasibility Assessment Conclusions Design Objectives and Performance Specifications Performance Specifications Nominal Speed Minimal Turning Radius Run Time Ground Clearance Acceleration Sensors Response Microcontroller Requirements Navigation Algorithm Design and Implementation Specifications Two Separate Drive Motors Belt Driven Four-Wheel Drive Gear motors Nickel-Metal Hydride Batteries Chassis Material: Aluminum Microcontroller Sensors H-bridge Evaluation Criteria Analysis Problem Statement Known Information Desired information Action taken Assumptions made Schematics and Data Hardware Design Description...39 MIMNS Interfacing...39 Sensor Interfacing... 40

3 H bridge interfacing Software Design Description Proposed Work/Schedule Milestones Schedule OPTIONAL (Not in scope) Appendix A...50 System Flow Chart List of Tables: Table 1: Table 2: Table 3: Table 4: Diagram 1: Diagram 2: Long List Short List Collective Average Feasibility Scoring (Electrical) Collective Average Feasibility Scoring (Mechanical) Basic Sensor Operation Timing Diagram (Obtained from DUR5200 datasheet.) H-bridge

4 4 1. Recognize and Quantify Need 1.1. Sponsor Background The Intelligence Community consists of agencies, bureaus, and organizations within the executive branch of the United States government that are critical to national intelligence. Such agencies include but are not limited to the Central Intelligence Agency, the Federal Bureau of Investigation, the National Security Agency, and the various military intelligence organizations. The Intelligence Community is responsible for the collection of information regarding organizations and activities that pose global and domestic concerns. This includes international terrorist and narcotics operations as well as intelligence or hostile activities directed against the US Project Mission Statement The Robotic Navigation Unit will be an independent vehicle, meaning it will carry all its own sensors, controller, and power supply, in addition to the MIMNS 1 cube. It will be approximately one square foot with a wheel diameter of approximately four inches. This design should be sufficiently robust for any terrain it may encounter while navigating the College of Engineering. The microcontroller will receive data from the MIMNS, as well as the infrared and sonar sensors. This information will be used to choose a direction for the robot, all the while recording the path information. If time allows, we will record the data from the sensors as well to create a more detailed map and make intelligent navigation decisions to ensure mapping of the entire accessible area. The robot will be driven by two independent motors, one for the right wheels and one for the left wheels, in order to provide the shortest turning radius possible. 1 See section 1.4.

5 Robot Navigation Background There are many lucrative applications in which an autonomous robot capable of mapping an inaccessible area would be valuable. Most robots in similar applications use an expensive video system or GPS to map an area of interest. However, there are instances where a complicated video system is too expensive or infeasible due to size constraints. GPS systems are more viable, but they cannot be used in areas without a clear view of the sky. Most robots for these applications must use video systems, a remote human, or a combination of the two. There is still a need for a simple yet effective method for mapping non-gps accessible areas without the dependence on a human. MIMNS offers an inexpensive alternative MIMNS MIMNS is a Miniature Inertial Measurement Navigation System developed last year at RIT as design project MD , based on the theory that signals from accelerometers can be used to estimate error in rate gyro signals and the ability to use signals from accelerometers and gyros to determine displacement. The MIMNS contains six accelerometers and three gyros, which output analog signal. Signal conditioning and interpretation via software can be implemented on the controller to accurately determine the displacement of the robot Scope Limitations The projected budget for the project should not exceed $2000 dollars. The prototype and final report shall be completed in a 20-week period and demonstrated to the sponsor by May This will include an autonomous unit capable of navigating a flat unknown and non-gps accessible area by utilizing the MIMNS cube. The unit must store its path based on the location determined by the MIMNS. The stored data must be able to be retrieved by a computer in a useable format for post processing. The vehicle must also be robust enough to survive several months of testing and provide repeatable results.

6 Stakeholders The primary stakeholder for this project is the Intelligence Community. A second stakeholder to this project is the Rochester Institute of Technology. Other possibilities include any other area where an autonomous robot is of possible use. Exploring abandoned mines, for example, is a task in which GPS control is not a viable option. However, this technology could also provide a cheap alternative to motion measurement devices currently employed in airplanes and other vehicles Key Business Goals The goal of this project is two-fold: to create a robot capable of exploring and mapping an unknown area, and to demonstrate the viability of the MIMNS cube as a motion measurement device. Most robots for exploring and mapping unknown areas use video systems, GPS or human control with some form of vision system. These methods all have certain weaknesses which could be overcome by using a system such as the Miniature Inertial Measurement /Navigation System (MIMNS) cube. Current positioning systems used in aircraft are bulky and require additional hardware for signal conditioning. This limits the vehicles (based on space capabilities) that can use the technology in addition to making the venture more expensive. A device similar to the MIMNS cube would be a cheap and smaller alternative to current devices Top Level Critical Financial Parameters The following parameters describe the critical financial parameters of the project: A budget of $2000 Production of one unit (capable of withstanding several months of testing) o Microcontroller. o MIMNS cube

7 7 o Chassis o Motors (DC and servo) o Sensors (sonar and infrared) o Batteries o Wheels o Belts and pulleys Field Testing 1.9. Financial Analysis The total budget for our project will be $2000. The microprocessor used is MSP430F449. A development kit, MSP430P440 (FET-Flash Emulation Tool), that contains the microprocessor as well as serial interface to connect to PC was purchased at a price of $50. This specific microcontroller was chosen both because of its processing capabilities and its memory allocation. The motors from Pittman Express, which cost $99 each, were chosen mainly because they had a built-in gearbox to eliminate an overly complicated belt system, decrease maintenance and increase longevity and reliability. The two H-bridges, which cost $30 each, were chosen because of their high current capabilities and easy to debug system. While sensors will not cost much individually, the total cost will be considerable. The batteries will likely be of the nickel-metal hydride variety, due to their low weight. We will use wheels, belts and pulleys similar to those commonly found on RC cars. There is no cost associated with the MIMNS cube, since we will be using the cube built by the previous senior design group Primary Market The primary market for the Robotic Navigation System is the Intelligence Community. It will be utilized to explore possibly dangerous areas inaccessible by standard GPS guided robots.

8 Secondary Market The secondary market for the Robotic Navigation System could be used or modified to suit many different applications. Bomb disposal robots, mine surveying robots and other robots could possibly incorporate a MIMNS motion system into their designs Teaser Topics What trade-offs are there to a faster/slower robot? - Robot should move at a speed at which it can collect information about its environment and process it in order to make decision which direction to set. Sensor response time sets one of speed limitations. Also, depending how long it takes to obtain accurate position information, the sampling frequency will either have to be decreased or the robot sped up or slowed down. Depending on the amount of information the robot is going to collect, the unit might have to make a full stop to collect information, process it and make a decision. If the robot is very slow, it will take unreasonably long time to map the area, keeping in mind that the robot will be battery operated and they have limited charge capacity. What trade-offs are there to a larger/smaller robot? - Considering that there was no limitation on the size or weight of the robot, the team chose dimensions according to the footprints of the components to be mounted on top of it. This robot will serve as testing unit and therefore its size is not very important. However, ultimately the size of the unit should be reduced to its minimum, but it should still be able to support environment sensing, navigation and mapping. What conditions is this robot capable of handling?

9 9 - The unit will be expected to operate in enclosed areas such as buildings and caves. Uneven terrain, as well as holes in the ground might be of consideration. Puddles pose a threat of damaging the electronics. Unit should be able to operate in the darkness. How would you approach the design differently for a harsher environment? - Chose components whose operation is not affected by temperature changes. Build protective cover. Possibly use tank design rather than wheels. Were there any noise issues that needed to be addressed? - To be determined Formal Statement of Work This team shall provide on fully functional and complete Robotic Navigation System that utilizes the MIMNS cube by May The robot will navigate its way through the second floor of the College of Engineering. It will record its position in the 2D space at regular intervals, and this will be used to create a map. This map must be transferable in a file format useable by a PC for post processing.

10 10 2. Concept Development 2.1. Long List When the group was assigned the Robotic Navigation Unit project the initial step was bringing up some general ideas. All ideas were documented, even the ones that were not considered afterwards. Some of the ideas are somewhat humorous as seen in the long list below in Table Group Drawings Since a picture is worth a thousand words, each member of the group sketched a picture of what the Robotic Navigation Unit should look like. During the work session the drawings were exchanged between the members and each member contributed and added to each of the drawings. Once all drawings have been reviewed, the group discussed the results and the long list was narrowed down Short List The sponsor of the Robotic Navigation Unit project is the CIA. The initial specifications from the CIA were the incorporation of the MIMNS (Miniature Inertial Measurement/Navigation System) cube, designed for project number MD last year, and the design of a non-gps autonomous system that would map all possible paths in an unknown area, while returning to its starting point. No restrictions were put on the size or the speed of the robot. For that reason, the physical shape of the robot was open for discussion. All shapes and materials shown in the long list in Table 1 were considered. The team eliminated the shark mobile, flying robot, and walking robot and kept the car-shaped robot, i.e. a rectangular shaped, and the circular shaped robot. As far as the composition is concerned, the team eliminated the use of wood, gold and platinum because of either cost or weight reasons, and kept potential use of aluminum or plastic, while keeping in mind that in the use of metal all circuit boards need to be covered to avoid shortening of the electronic components. As far as the electronic ideas, it was decided to use either a one or two microcontrollers, depending on the complexity of the other components of the robot such as the sensors. The type of the microcontroller will be determined further on into the

11 11 project design. As far as the sensors are concerned, the selection was narrowed to using ultrasonic sensors and IR sensors. The number of the sensors was to be determined later on. Laser sensors were eliminated because of the high cost (~$3000 and up), and the photoelectric sensor does not apply to the purpose of mapping caves. As far as the memory devices are concerned, the selection was narrowed down to either using only the memory present in the microcontroller or adding external RAM or Flash disks. The name for the robot will be determined also. Suggestions have not been made for that yet. Table 2 below summarizes the short list items. Shape Composition Electronics Sensors Memory Miscellaneous Table 1: Long List Shape Composition Electronics Car Truck Race car Shark mobile Walking robot Flying robot Circular Aluminum Plastic Wood Single bread board Each circuit on different board Digital display Single microcontroller Two microcontrollers Ultra-sonic sensors IR sensors Laser sensors Photo-electric sensors Fiber optics Rely on microcontrollers Internal memory only Additional external memory: RAM, Prom, Eprom, Flash Name for robot Water resistance Self destruction capability Car Circular Aluminum Plastic Single bread board Single microcontroller Two microcontrollers

12 12 Sensors Memory Miscellaneous Ultra-sonic sensors IR sensors Rely on microcontrollers Internal memory only Additional external memory: RAM, Flash Name for robot Table 2: Short List

13 13 3 Feasibility Assessment (Electrical) 3.1 Feasibility Strategy During the concept development phase, the team determined that the major electronic portion of the Robotic Navigation project has to consist of at least one microcontroller. Several were considered: the 8051 Intel, Oopic and MSP430 (a single MSP- FET430P440 was chosen after a careful analysis). The group also considered various types and combinations of sensors: ultrasonic sensors, laser sensors and infrared sensors (ultimately a combination of three ultrasonic and two infrared sensors were chosen). The dimensions of the microcontroller and sensors were not critical for this project. The three conceptual designs are similar except for the different components and different vendors. Also one of the concepts includes two microcontrollers communicating with each other. We have completed the Project Feasibility Assessment Worksheet, Estimation of Relative Importance of Attributes Worksheet and Weighted Evaluation Worksheet. From the first worksheet, we have compared the sufficient skills, sufficient equipment, sufficient number of people, percent of total required funds we have, chances of meeting the intermediate mileposts, chances of meeting the PDR requirements, chances of meeting the CDR requirements and overall technical feasibility of the project. We felt confident about our skills and we thought we would learn as we progress with the project. Also the same rating was assigned to the chances of meeting the intermediate mileposts and the technical feasibility. At the time we could not predict what would happen at the end of the quarter. We felt confident or strongly confident about the sufficient equipment (enough resources to purchase whatever necessary for the project), sufficient number of people and our funding was also sufficient. We also predicted that we could meet the two most important deadlines of the Preliminary Design Review (PDR) and the Critical Design Review (CDR). According to the Estimation of Relative Importance of Attributes Worksheet, we assigned a high value to meeting the PDR and the CDR requirements. In the Weighted Evaluation Worksheet, three concepts in addition to the baseline concept were selected. The Baseline Concept was to use the MSP430 microcontroller

14 14 with the three ultra sonic sensors. The first concept was to use the 8051 Intel microcontroller with three ultra sonic sensors. The second concept proposed was the MSP430 with six laser sensors and finally the third was the MSP430 with three ultra sonic sensors and the two IR sensors. After detailed analysis, we came to conclusion that concept three theoretically performed the best. We felt that we had enough skills to attack the problem of integrating the MSP430 with the five sensors even though there would be a learning curve encountered in the process. Also with the research done by the members of the group, we came to the conclusion that the selected microcontroller had more than enough functions needed to solve the problem. The next step in the formal feasibility assessment was to answer some general questions regarding the technical, economic, market, scheduling, and performance feasibility of each design. Each design was assigned a rating between 0 and 3 for each of the 10 questions. This rating represents the feasibility of the design in question with respect to the issues raised by a particular question. A rating of 0 represents very low feasibility, a 1 signifies that the design in question is slightly worse than the baseline concept, a 2 indicates that the design is approximately the same as the baseline concept and a rating of 3 signifies that the design is an improvement over the baseline concept. All team members completed this rating process, and the responses were averaged. The following section presents the questions by which the concept was judged, as well as explanations of the rationale behind the ratings assigned by the team. 3.2 Feasibility Questions Technical Question 1: Does the team have the skills needed to implement all aspects of the technologies for this concept? Rating (Concept1 (C1)=3, Concept2 (C2)=2, Concept3 (C3)=3) Concepts 1 and 3 were equally difficult in implementing, however both scored higher than implementing six sensors. We felt that as a group we could tackle the problem of integrating either of the two microcontrollers with the specified sensors. Technical Question 2: Are all elements of the technology required to implement this concept available to the project team?

15 15 Rating (C1=2, C2=1, C3=3) All the components are readily available from the vendors for purchase, however the 8051 Intel microcontroller is an 8-bit microcontroller in comparison to the 16-bit MSP430 and also the cost of the laser sensors is too high. Our group will be creating a purchase requisition form in order to obtain the necessary components. We are ordering all our components except for the Miniature Inertial Measurement/Navigation System cube and then integrating all of them to communicate with each other. Economic Question 1: Do we have enough funds to support the development of this product? Rating (C1=2, C2=1, C3=3) The first concept is more expensive due to higher cost for the 8051 development tools. The second concept was out of the question due to high cost of the laser sensors (starting at ~ $3000 each). The third concept meets both the cost and functionality requirements. Economic Question 2: Does this product appear to have viability in the future economy? Rating (C1=3, C2=1, C3=3) Any of the three concepts has a potential in the future economy. The third concept however is the least expensive and the least complicated from our standpoint. The demand for such a product will only increase with such improvements presented above. Market Question 1: What price will the market bear for a product based on this concept? Can we afford to make and distribute the product at that price? Rating (C1=3, C2=1, C3=3) We expect the demand for this product to be the same for the first and the third concept and a lower demand for the concept 2 because of the high cost of the laser sensors. The non-gps Navigational Unit with the MIMNS cube will offer a more convenient and less expensive navigation and obstacle avoidance solution than other systems with comparable accuracy.

16 16 Market Question 2: How does this product fit with our current and future areas of strength? Rating (C1=2, C2=2, C3=3) Due to customer requirements and specifications the third concept is very similar to the baseline concept except for the addition of the two IR sensors for greater accuracy of seeing an obstacle. Schedule Question 1: How much time will it take to bring the concept to the customer? Does our firm have the time to bring this product to the market? Rating (C1=3, C2=3, C3=3) There are no constraints for this project for the size, improved accuracy, and minimal cost. This is essentially a proof of concept in continuation of a project from last year, MD , to illustrate that in fact the MIMNS can work in the field. Schedule Question 2: How long is the window of opportunity in the market? Rating (C1=3, C2=2, C3=3) The need for a product such as this is not limited to a certain widow of opportunity. The product will introduce a unique solution to a preexisting problem and can be integrated into the market when it is completed. Performance Question 1: How does this concept meet the top three needs of our immediate customer for this product? Rating (C1=1, C2=0, C3=3) The top three needs of the CIA are for this prototype to be under $2000, to be able to navigate and avoid obstacles without the use of GPS and to communicate and record data from the MIMNS cube. Performance Question 2: Can this concept satisfy the needs of an additional user beyond those of our baseline customer that it is being designed for? Rating (C1=3,C2=2, C3=3)

17 17 The product is easy adaptable and could be implemented into any number of applications. The baseline customer is only the initial application of such a product. The Non-GPS Robot Navigation System is a revolutionary idea with the capability to significantly influence the market. T1 T2 E1 E2 M1 M2 S1 S2 P1 P2 Concept Concept Concept Table 3: Collective Average Feasibility Scoring 3.3 Initial Feasibility Assessment Conclusions The conclusion was drawn from the above table and the various analyses completed. Concept 3 clearly is the best solution for the Non-GPS Robot Navigation System.

18 18 4. Feasibility Assessment (Mechanical) Feasibility Strategy The Mechanical portion of the concept development produced three concepts for the robot chassis. These three concepts could be used for any set up of electronic equipment giving our group the best range of options to produce a more effective overall design. The first design was a flat rectangular plate with walls to house the electronic and mechanical components, protecting them and giving the chassis rigidity. The second design was just a flat rectangular plate for the chassis. The components then would be bolted and bracketed to the plate, allowing freedom of placement and adjustment, but lacking some stability. A round plate with no walls was considered for the third concept. This plate would be like the rectangular plate in allowing freedom for adjustment and defining a general location for the sensors. The third design also complicates motor and power transmission placement, along complicating with wheel locations. The three concepts are weighted in a Weighted Evaluation Worksheet, in which we came to a conclusion on a mechanical design for the robot. Based on collective research and some quick layouts of the concepts, it was noted that design concepts 2 and 3 were easy, cheap, and effective, while design 1 was strong, cheap, effective, but more difficult to execute Feasibility Questions Technical Question 1: Does the team have the skills needed to implement all aspects of the technologies for this concept? Rating (Concept1 (C1)= 1, Concept2 (C2)= 3, Concept3 (C3)= 2) All 3 concepts are easily implemented, but concepts 2 and 2 were a more robust design and much more easily modified. The first concept has an advantage over the second in that it has more rigidity based on the hull like structure; unfortunately this also caused interference problems with shafts and the motors. Since the first concept had such a major flaw, the second design was considered a simpler and more modifiable design.

19 19 The third concept has the same characteristic in adjustability as the second concept except that it is round. The roundness of concept three creates difficulties when aligning wheels and motors properly. Technical Question 2: Are all elements of the technology required to implement this concept available to the project team? Rating (C1= 2, C2= 2, C3= 2) All materials and components are easily obtained through various vendors. The team has decided on using a sheet of aluminum for the chassis of the vehicle for all the concepts. Aluminum was chosen for its material characteristics of being very light, very strong and easily machined. The other components such as bearings, gears, motors and batteries are obtained through various Hobbyist and RC/car vendors. All three concepts obtained the base average on this since all concepts would use similar mechanical components. Economic Question 1: Do we have enough funds to support the development of this product? Rating (C1=2, C2=2, C3=2) With the exception of the MIMNS cube, which is one of a kind at the moment, the most expensive mechanical components would be the motors, ranging from $30 each to upwards of $150 each. Considering the rest of the mechanical components are fairly cheap, most on the order of $30 and the few electronic components there will be plenty of funds to support the product development. Economic Question 2: Does this product appear to have viability in the future economy? Rating (C1=2, C2=2, C3=2) Considering that this project is not for the consumer and most likely not for industry the viability lies in governmental use. While the robot is for government use it also needs to be cost effective. The overall idea is to have many of these robots scouting out an area at once, so the cost must be low to allow a mass production of the robots. All of the current concepts will roughly cost the same.

20 20 Market Question 1: What price will the market bear for a product based on this concept? Can we afford to make and distribute the product at that price? Rating (C1=2, C2=2, C3=2) Since all the concepts have the same majority of parts the cost for the mechanical system of the robots will be about the same. One of the expressed needs was that this robot had to be cheap, so there can be a sort of mass production at some point. Even though this isn t really for the market, it will be able to be marketed if there becomes a need. Market Question 2: How does this product fit with our current and future areas of strength? Rating (C1=2, C2=2, C3=2) Currently there are similar type robots and measurement technology. But this robot will be the first to combine the two. Since all robots will have the same mechanical technology they will be equal in that respect. Schedule Question 1: How much time will it take to bring the concept to the customer? Does our firm have the time to bring this product to the market? Rating (C1=3, C2=3, C3=3) Since this is not a final product it will be a while before the completed product will be brought to the customer. This portion will however be finished before the time allotted including the debugging stages. Schedule Question 2: How long is the window of opportunity in the market? Rating (C1=2, C2=2, C3=2) This portion of the product has a large window of opportunity. It is designed so it can be used in other projects as well as the next phases of this project.

21 21 Performance Question 1: How does this concept meet the top three needs of our immediate customer for this product? Rating (C1=2, C2=2, C3=2) The robot is large enough to hold all the components safely and is small enough to be useful, for example it can maneuver through doorways. The mechanical portion is also relative low cost allowing for better sensors and electronics and increasing functionality. Performance Question 2: Can this concept satisfy the needs of an additional user beyond those of our baseline customer that it is being designed for? Rating (C1=2,C2=2, C3=2) As said before this robot can be used for anything people see necessary. The mechanical portion is easily modified to fit a number of other uses. T1 T2 E1 E2 M1 M2 S1 S2 P1 P2 Concept Concept Concept Table 4: Collective Average Feasibility Scoring Initial Feasibility Assessment Conclusions The conclusion can be drawn from the above table and the various analyses completed. Concept 3 clearly is the best solution for the Non-GPS Robot Navigation System.

22 22 5. Design Objectives and Performance Specifications 5.1.Performance Specifications The following section explains how the proposed solution meets the design objectives and performance requirements based on considerations explained in the needs assessment and the feasibility assessment. Design objectives: Integration of the Miniature Inertial Navigation System (MIMNS) developed in a previous senior design project Development and integration of an autonomous routing control system Construction of a robot with "intelligent" capability demonstrating the mapping of an enclosed environment Performance Questions Evaluated: Size and weight of the unit? Speed of the unit? Sensor response? MIMNS interfacing? Obstacle avoidance method? Navigation? Mapping? Cost?

23 Nominal Speed A minimum speed was not specified by the sponsor; however, the design team decided that a nominal cruising speed should be set in order to determine the performance requirements of the various components. It was important that the robot goes slow enough so that it could stop or turn when an object was detected in its path. Its speed is limited by the sensor response time, as well as the time which takes microprocessor to convert MIMNS data into position coordinates. The design team decided that the robot should be capable of navigating and mapping the second floor of the College of Engineering in less than 30 minutes. Factoring in pauses to take sensor readings of the surrounding environment, the robot s required cruising speed was estimated at 4 mph, or about 5.9 ft/s (1.8 m/s). Nominal Speed, v = 5.9 ft/s (1.8 m/s) Minimal Turning Radius Because the robot may be operated in narrow tunnels or hallways, maneuverability is an important factor in its overall performance. In order to achieve the greatest maneuverability it was agreed that the turning radius needed to be kept as small as possible. Therefore the robot must be able to rotate in place about its centerline Run Time To allow for the navigation and mapping of large areas it was decided that the robot must be able to operate for at least 1 hour between battery charges. This extended run time will also be beneficial during the testing phase because it will reduce downtime due to battery charging Ground Clearance Although this non-gps mapping robot will not be subject to harsh or bumpy terrain, it must have adequate ground clearance to easily traverse door jambs and other small obstacles it may encounter while navigating the second floor of the engineering building. As a result, the minimum ground clearance was set at 1 inch.

24 Acceleration While the robot s acceleration is not of critical importance in terms of performance, the desired acceleration will help in determining the required torque output of the motors. The design team agreed on an acceleration of approximately 3 ft/s 2 (~1 m/s 2 ). This acceleration requirement is low enough to allow the use of fairly small drive motors but will also enable the robot to reach its nominal cruising speed within 2.5 seconds Sensors Response Sensors should be able to detect obstacles up to 3.4 meters and pass reliable information in regards to their location back to the microcontroller. Detailed mapping is not the main objective of this project and accuracy of the distance between the unit and the obstacle is not of major concern. The sensor response time (in case no obstacle is detected) should be fast enough to give robot time to decide which is the proper direction to take, without stopping. Initially, only one sensor on the servo will be used for obstacle detection. In the second phase of the project, additional sensors will be mounted on the side to provide microprocessor with additional information. The sensor will communicate to the microcontroller and therefore should preferably have digital interface. It should not have power requirement higher than 5V Microcontroller Requirements Microcontroller will be used for gathering and processing of the information. It needs to collect information from the sensors, obtain distance from it so it can make a judgment which way to turn. It should utilize MIMNS data as reference to where it is and where it is going, i.e. in two dimensions. The same data will be used to store information about the path taken. When it becomes possible to retrieve accurate data about the path taken from the MIMNS, it will not be much more difficult to obtain a more detailed map about the robot s surroundings by utilizing and storing information coming from the sensors. Additionally, the processor needs to control the two D.C. motors chosen as the drive mechanism.

25 25 In order to cover all of the above requirements, it was decided that the unit shall have I/O ports with interrupt capability, timers, A/D converter, accurate clocking system and low power consumption (battery operated). Additionally, a development kit should be available which will have a C compiler considering that the team does not have much experience with assembly programming Navigation Algorithm A simple obstacle avoidance algorithm, in addition to gathering information from the MIMNS is to be implemented first in the design development. When the robot detects an obstacle it will look to the left and the right of it, take measurements and decide which way is the safest to go (furthest object or no object detected). Secondly, the unit will utilize information about its position provided by MIMNS to ensure it keeps going into an unknown area, i.e. give intelligence to the unit in its decision making in addition to the obstacle avoidance. Finally a smarter algorithm will be attempted for the second stage implementation. This algorithm will ensure the unit covers the entire floor and returns to its starting point. The flowchart shows the operation of the unit in Appendix A. 5.2.Design and Implementation Specifications Most of the performance specifications were set in order to accomplish the previously stated design goal. Considering that a there were no specific limitations on the size, the weight of the robot and there was no speed requirement; the specs were set to simplify the design or fabrication of the robot Two Separate Drive Motors In order to accomplish the performance specification of minimizing the robot s turning radius, a drive train was selected in which each side of the vehicle is controlled and driven independently. This design will allow the robot to pivot in place by driving the wheels on one side in one direction and the wheels on the other side in the opposite direction. This layout also simplifies the design of the robot because it can be steered

26 26 without the use of an automotive style steering system in which the front or rear (or both) wheels are turned left or right Belt Driven Four-Wheel Drive A belt drive system was selected to transfer power from the motor on each side to the wheels on each side. This four-wheel drive system is important in minimizing the turn radius, because it will allow the robot to rotate about its centerline whereas a twowheel drive system would only allow the robot to rotate about the midpoint of the driven axle. Timing belts and pulleys were chosen for their relative simplicity and lower cost when compared to shaft or chain drives Gear motors The team decided to use gear motors (motors with integral gear-reduction) in the design mainly to simplify the drive train. Because the required wheel speed is on the order of 400 rpm and most motors run at speeds on the order of 5000 rpm, a significant gear-reduction is required to obtain the desired nominal cruising speed. Incorporating this gear-reduction into the belt drive system would add complexity to the drive train. Using a gear motor with an output speed close to the required wheel speed allows the use of a simple belt drive system with a gear ratio at or near 1: Nickel-Metal Hydride Batteries Due of the amount of testing planned for the robot, rechargeable batteries will be required. Nickel-Metal Hydride (Ni-MH) batteries were selected to power the robot, because of their increased capacity when compared to other types of rechargeable batteries such as Nickel-Cadmium (Ni-Cd). Another benefit of Ni-MH batteries is the fact that they have no memory effect in which their performance decreases if not completely discharged before charging. Ni-MH batteries are considerably less expensive than the newer Lithium-Ion rechargeable batteries and are also better suited for higher current draw applications.

27 Chassis Material: Aluminum Aluminum was selected for the construction of the robot s major chassis components for many reasons. Primarily, aluminum will provide the chassis far greater strength and stiffness than plastic while keeping its weight relatively light. The use of aluminum will also ease machining when compared to steel or plastic materials Microcontroller Texas Instruments MSP430 is a chosen microcontroller for this project, specifically, MSP430P40 Flash Emulation Tool. It has six 8-bit I/O ports, internal 12bit 8-channel A/D converter and 60k programmable Flash, with serial communication interface (USART). The controller also has a clock system which includes Hz watch crystal oscillator, internal digitally controlled oscillator (DCO) and high frequency crystal oscillator. Additionally there is an integrated LCD driver for displays up to 160 segments which is not needed but might be useful in the testing phase. Three 16 bit timers with 7 capture and compare registers and the built in comparator will probably be utilized to obtain information from the ultrasonic sensor(s). Part of the development kit is a C-compiler limited to 2k of code. 60kB of memory is sufficient to store information about the path taken (record x and y coordinates four times per second) and possibly store a more detailed map which will be created if time allows. MSP430 was also chosen because of its low cost; $50 for the FET Development Kit which includes JTAG connector, crystal oscillator a two MSP430F449 Microprocessor ICs Sensors Two types of sensors were chosen during the design. The first is ultrasonic sensor and the second is infrared sensor. The ultrasonic sensor will be responsible for obstacle avoidance and navigation. They send an out a sonar burst and an echo is returned to the microcontoller carrying distance information. The echo pulse is traveling at the speed of sound (~343m/s), and in order to allow long enough processing time it was decided to trigger the ultrasonic sensors not more often than every 50ms (which is the sensor maximum response time, in case no obstacle was detected). The ultra sonic sensor chosen

28 28 was DUR5200. DUR5200 was the one the team chose because it has digital output which enables direct interaction with the I/O pins of the controller. The sensor has a relatively low voltage supply of 5V and a range of 4cm-340 cm. Cost was also taken into consideration and since the cost of these sensors was relatively low (~$30 each) and for all the reasons above, the DUR5200 was chosen. One ultrasonic sensor will be placed on top of a servomotor in the front (middle) part of the robot and will supply information back to the microcontroller with respect to a safe path ahead of the robot. In case of obstacle detection, the sensor will be turned 90 to obtain additional paths information and the decision will be made by the microcontroller. Two additional ultrasonic sensors will be mounted to the left and right sides on the robot and the information taken from the two sensors will be utilized in stage two of the design, explained in the sections above and allow a more detailed mapping of the area. This will shorten the decision making time as for which path to take next. Two infrared sensors will be mounted on the front (edges) of the robot to allow immediate obstacle detection. The infrared sensors send a light wave and their range is shorter than that of the ultrasonic, though more accurate. The range is up to 40cm. When the IR sensors sense an obstacle, the robot will stop immediately and search for additional paths. This is to allow a more accurate obstacle avoidance technique. The IR sensors chosen are the Sharp GP2D12. These sensors are highly reliable and use the same voltage supply as the ultrasonic ones, i.e. 5V. They are used in numerous autonomous systems and are relatively cheap (~$12 each). During the testing stage, if necessary, additional sensors will be mounted H-bridge The H-bridge acts as a switch between microcontroller and the two D.C. motors. It connects to the microcontroller per two pins. Depending on the polarity (high or low) of the pins, forward or reverse motion of the motor can be accomplished and by modulating the pulse width, different speed settings are possible. The H-bridge connects to the motor per two other pins switching of which also accomplishes forward and reverse motion. However, only software control is considered. The H-bridge also provides terminal for 12 Volt battery connection needed for the D.C. Motors.

29 Evaluation Criteria Considering mechanical aspect of the project, it can be easily measured whether the performance specifications are met. The nominal speed can be calculated by recording the time required to travel a known distance. The turning radius and ground clearance can be measured with a tape measure. The robot s run time can be easily measured with a stopwatch during testing. The robot s acceleration may be more difficult to accurately measure, but it could be estimated by measuring the amount of time required to accelerate to its nominal cruising speed from a stop. Because the acceleration is not of critical importance, this estimate should be enough information for an evaluation of performance. The electrical part is relatively straightforward as well. The microprocessor is capable of generating a controlled pulse waveform, which will feed directly into the H- bridge, which connects with two pins to the motors, additionally it provides power connection for the D.C. Motors (12V). The sensor requires a trigger signal sent out from the microcontroller as well as an echo pin, which will sense the rising edge of the returned signal. It also connects directly to the microcontroller with possibility that a low voltage regulator might be needed considering that the microcontroller voltage is max 3.9V and the level of the signal coming from the sensor is 5V. The MIMNS is outputting six analog waveforms that feed into the internal 12bit 8-channel A/D converter. Programming microcontroller and implementing the navigation algorithm will be the most challenging part of the project.

30 30 6. Analysis After considering all the objective and performance specifications several issues came up which needed to be resolved before a system could be proposed which would be able to meet all of the specification previously discussed. This section discusses these issues and proposed solutions. At the end a system schematic is presented as well as navigation flowchart which will be implemented in phase 1 (the required phase) of the project. 6.1.Problem Statement a) Problem 1: What is the required angular velocity of the wheels? To attain the desired nominal cruising speed of the robot the wheels must turn at a specific angular velocity. b) Problem 2: What is the required gear ratio between the motor and wheels? Because the motor output speed is relatively fixed at a known value and the wheels are required to turn at the specific angular velocity calculated in Problem 1 a gear ratio may be required to obtain the desired wheel speed. c) Problem 3: What is the minimum required torque output from each motor? The motors must be capable of supplying enough torque to accelerate the vehicle at the desired rate of acceleration. To ensure that the motors chosen will be powerful enough the minimum torque required must be calculated. d) Problem 4: What is the minimum required tensile strength of the drive belts? To ensure that the drive belts will be capable of transferring the required torque without failing the maximum tension that the belts will see must be determined. e) Problem 5: Robot position needs to be recorded.

31 31 In order for the unit to create a map, it needs to be aware of its position and direction. Considering that the unit will cover non-gps areas, autonomous means need to be found. f) Problem 6: Robot needs to be aware of its surrounding. In order for the unit to successfully avoid obstacles or make decisions which is the proper path to take, it needs to be able to see its surroundings and analyze the data taken. The sensor response time should be fast enough for the given speed of the robot to pass information about obstacles. The readings should be reliable as far as obstacle detection, and obstacle identifications regardless of their size. The sensors should also not interfere with each other s operation. They need to be placed on the robot in such a way to cover up to 180 degrees vision range. g) Problem 7: Robot needs to have full control of its motion/direction/turning. The robot should have capability to move without external, remote control. The commands of the motion direction will be passed by the microcontroller and the robot should have means to convert that electrical information into mechanical motion. The motion/turning should be precise and if possible the unit should be able to move at different speeds. h) Problem 8: Robot needs to store the complete path taken. The robot should be able to maintain information about the path taken, possibly a map of its surrounding as well. This data should be in binary form, easily passed to the PC which will use it to create a visual map which represents the path/map of the area covered by the unit. At this accuracy is not major issue, but rather capability of the unit to utilize that data in its decision making process.

32 Known Information a) Problem 1: What is the required angular velocity of the wheels? The desired nominal speed of the robot is 1.8 m/s (5.9 ft/s). The wheel radius is approximately 57 mm (2.25 in). b) Problem 2: What is the required gear ratio between the motor and wheels? The output speed of the motor is approximately 420 rpm. The desired wheel speed is 300 rpm. c) Problem 3: What is the minimum required torque output from each motor? The gear ratio between the motor and the wheels is approximately 1.4:1. The desired acceleration of the robot is 1.8 m/s 2 (5.9 ft/s 2 ). Assume the total mass of the robot is approximately 5 kg (11 lb) d) Problem 4: What is the minimum required tensile strength of the drive belts? The pitch radius of the motor pulley is 7.77 mm (0.306 in). The torque on each motor pulley is approximately 91.8 N*mm (13 oz*in) e) Problem 5: Robot position needs to be recorded. The MIMNS cube (Miniature Inertial Measurement Navigation System) passes six analog waveforms containing acceleration information, to the microcontroller s internal, 8-channel ADC. An algorithm will be implemented to process this data, i.e. integrate the acceleration twice and turn it into position. The position will be recorded into the Flash memory. Microcontoller has interrupt capability as well as several counters which can be used to trigger A/D conversion at specific time intervals and pass the data into some register in the memory. There should be enough memory to store information about the path. f) Problem 6: Robot needs to be aware of its surrounding. Sensors chosen for this project are the DUR5200, which are ultrasonic sensors with a detection range of 4-340cm as well as Sharp GP2D12 Infrared sensors with

33 33 a detection range of up to 40cm. The IR sensors will be used for immediate obstacle detection where the ultrasonic sensors will be used for navigation and obstacle avoidance. g) Problem 7: Robot needs to have full control of its motion/direction/turning. H-bridge is used to interface between the microcontroller and the two D.C. motors. It receives command from the microcontroller per two pins and depending on the polarity of the pins, forward as well as reverse motion are achieved. Additionally, by passing a pulse waveform with adjustable duty cycle to the H-bridge, it is possible to achieve forward and reverse motion of the motor. The H-bridge connects to the D.C.motor per two more pins (reversal of which can also achieve forward/reverse motion) and provides terminal for the 12 Volt battery needed by the D.C. motor. The H-bridge itself is requires about 4V power. h) Problem 8: Robot needs to store the complete path taken. The mircoprocessor has internal, 60kB of (programmable) Flash memory which should be sufficient to store relatively detailed information about the path taken by the robot assuming the unit Desired information a) Problem 1: What is the required angular velocity of the wheels? The angular velocity of the wheels required to obtain the desired robot speed must be determined. b) Problem 2: What is the required gear ratio between the motor and wheels? The gear ratio between the motor and the wheels required to obtain the desired wheel speed must be calculated. c) Problem 3: What is the minimum required torque output from each motor? The minimum torque required from each motor in able to accelerate the robot at

34 34 the desired rate of acceleration must be determined in order to select motors with adequate power output. d) Problem 4: What is the minimum required tensile strength of the drive belts? The maximum tension found in each belt must be calculated to ensure that the belts selected will not fail. e) Problem 5: Robot position needs to be recorded. A/D conversion of the MIMNS data needs to be tested. It is desired to find out exactly how much time it will take for the processor to execute the algorithm which will give the position information. Additionally, the accuracy of this information, as well as what it affecting it (if anything) should be known. f) Problem 6: Robot needs to be aware of its surrounding. Sensors need to be thoroughly tested before they get integrated in the whole system. Even though datasheet for the sensors contains information about the range covered as well as microcontroller interfacing, this cannot necessarily be trusted. It is necessary to find out exactly what range the robot can see as well as what it takes to pass that information back to microcontroller. Once this process is clear, it won t be too difficult to decide on the exact placement of the sensors on the unit and to integrate sensor information in the overall navigation/decision making algorithm g) Problem 7: Robot needs to have full control of its motion/direction/turning. Operation of the H-bridge if fairly straight forward and the component is in phase of testing right now. The forward and reverse motion has been tested by using lab equipment, power supply and a function generator capable of supplying pulse waveforms with different period, duty cycle etc. h) Problem 8: Robot needs to store the complete path taken. The information about the path covered needs to be written into memory (in order

35 35 to create a map) as well as read from it (in order to be utilized in decision making process). It is necessary to investigate exactly how this can be done in the microcontroller. 6.2.Action taken Assumptions made a) Assuming that the wheels are rolling without slipping the angular velocity required to produce the desired robot nominal speed is easily calculated: ω V = 1.8 m/s r = 57 mm V r m 1.8 V s r 0.057m 300rpm rad s b) Assuming that the motor output speed is fixed at 420 rpm the gear ratio required to turn the wheels at 300 rpm can be determined:

36 36 Motor Pulley Wheel Pulley r 1 r 2 ω rpm ω rpm r r r r2 r r 2 300rpm 420rpm c) Assuming the robot has a mass of approximately 5 kg (11 lb) and that the desired rate of acceleration is 1.8 m/s 2 (5.9 ft/s 2 ) the required torque output from each motor can be calculated: The tractive force required at the wheel/ground interface must first be found: m F m * a 5kg *1.8 2 s F 9N The tractive force, F Wheel, required at each wheel is equal to F/4 = 2.25 N. The torque required at each wheel can then be determined: T Wheel = F Wheel * r = 2.25 N * 57 mm

37 37 T Wheel = N*mm (18.16 oz*in) The torque at each motor pulley can be calculated using the gear ratio between the motor and the wheels: T T MotorPulley MotorPulley T Wheel r * r MotorPulley WheelPulley 91.61N * mm(12.97oz * in) (128.25N * mm)* Because each motor drives two motor pulleys the total torque required from each motor is equal to twice T MotorPulley : T Motor = 2 * T MotorPulley T Motor = N*mm (25.95 oz*in) d) Assuming the pitch radius of the motor pulley is 7.77 mm (0.306 in) the tension in each belt can be calculated using the known value of T MotorPulley : T MotorPulley F F Belt Belt T r F Belt MotorPulley MotorPulley * r MotorPulley 11.79N (2.65lb) 91.61N * mm 7.77mm e) Position data should be sampled independent of what else is going on with the controls (sensor information processing, H-bridge control etc.) Additionally, it should not affect the controls of the system, avoiding obstacles being one of the priorities. The position information will be stored in a byte per set of coordinate points. That way, it will be possible to store more than data points. f) Three ultrasonic sensors and two IR sensors should make up the sensing part of the controls. Main focus will be on the sensor which will be placed in the middle of the unit, up front, mounted on the servo and have the capability of covering up to 180 degrees range. In the first phase of the project only that sensor will be used

38 38 for obstacle avoidance. The IR sensors serve rather as bumping protection and ensure that even in case ultrasonic sensors fail, the unit won t bump into anything. In the second phase of the project, two additional sensors will be added to the left and right side of the robot to provide additional information and reduce the turning radius (time) of the sensor on the servo. g) The microcontroller will be able to generate a pulse waveform to drive the H- Bridge with adjustable duty cycle so it can control the speed of the motor in addition to the motion direction of the robot. h) A learning curve will be necessary in order to get familiar with microcontroller s operations, its different features and to master programming Schematics and Data

39 39 Trigger TE RS TE RS TE RS Echo I/O Ports Capture/Compare register 00 Fwd 01 Left 10 Right 11 = Back MSP430 Processor I/O Pins / PWM generator Memory (Flash) 8 Channel 12 bit ADC IR3220S H-Bridge IR3220S H-Bridge MIMNS IR IR DC Motor 1 DC Motor Hardware Design Description System schematic, in Appendix A, shows system schematic. The arrows represent direction of the data flow. As can be seen in the diagram, microprocessor is the main control unit and is talking to all other subsystems. It requires not more than 3.9V power. Batteries (possibly two regular 1.5V AA) will be used to power it up. Microcontroller has to cover three basic tasks: get position information from MIMNS, get environment information from the sensors and pass motion command to the H-bridge. MIMNS Interfacing - As mentioned earlier, MSP430F440, processor chosen for this project, has internal 12 bits, 8-channel A/D converter with internal reference. Output of the MIMNS cube (6 analog waveforms containing acceleration data) feeds directly into the A/D port on the processor. A software interrupt will

40 40 control ADC reading frequency, so that A/D converter can be read periodically, regardless what else is going on in the system. No information is send from the processor to the MIMNS. The time it takes the processor to obtain accurate position information from the data passed to it by MIMNS, might be a limitation on the speed of operation. The data from the A/D converter as well as coordinate points will be stored in the memory. The processor has 60kB internal Flash memory which is also used as the program memory. This should be enough for initial testing of the unit and controls system. Depending on the desired accuracy of the map (more data points or even more bytes allocated per data point) this amount of memory might not be sufficient. Sensor Interfacing - The ultrasonic sensor chosen for this project requires 5V and ground connection to operate. It is controlled by the microcontroller per two pins (TE transmit enable and RS-receive signal). After the controller sends out a 10u sec (t1, see diagram below) trigger signal, the sensor turns on, sends out the 40kHz, ultrasonic burst (generated on the sensor board) and than pass the echo pulse back to the controller. This information is than processed into distance measurement between the robot and the obstacles, which is used by software to make decision which way the robot should go. The formula used to determine distance to the object (in m) is: v( m / 2) Dis tan ce( m) td (sec) 2 Td is the time delay between triggering the ultrasonic burst and sensing high on the echo pin.

41 41 Diagram 1: Basic Sensor Operation Timing Diagram (Obtained from DUR5200 datasheet.) The solution to measuring the exact time delay is to use one of the timers with input capture register whose function can record the time at which an external even takes place and latch the content of the timer register at that time. This value can then be used to obtain time delay. A generic I/O pin can be used to set the trigger. Considering that the sensor operates of 5V and the signal it s sending to the microcontroller will most likely be at that level (still needs to be confirmed), it might be necessary to add a series resistance to prevent latch up or excess input current. H bridge interfacing - H-bridge is also controlled per two pins by the microcontroller. It takes commands from it (high and low) and does not pass anything back. It will share the same battery set with the D.C. Motors, however requires only 5V. Therefore a low voltage controller LM7815 will be used to generate the required voltage from the 12V battery required for the D.C. Motor. The operation of the H-bridge is shown below.

42 42 Diagram #2: H-bridge The values in the truth table represent the logic levels at the microcontroller pins. Initially, the idea was to use general I/O pins from the micro to control the H-bridge, but after some basic testing, the team realized that the speed of the motor can be controlled easily by varying the duty cycle of the pulse wave feeding into one of the pins. This caused us to reconsider the approach and use a Timer port to generate pulse waveform, write up a function which will modify the duty cycle under given conditions (no obstacle-go faster, obstacle-go slower) and vary the polarity of the second pin, depending whether forward or reverse motion is required. Testing phase will include the maximum current on the I/O pins in active mode and confirm that it is enough to drive the H-bridge. In case the H- bridge requires ore current than that, a transistor with current gain ( Darlington Sink Driver, ULN2803AP) will have to be added between micro pin and the H-bridge Software Design Description Software development is divided into four major stages. The scope of this project requires only completion of the first two stages, the other two have been made optional. Stage 1.

43 43 Study microcontroller operation and its features. Become familiar with the simulator used in the development kit. Test the microcontroller using LED s, oscilloscope, voltage/current meters. Test ADC. Understand how memory is written to and how information is retrieved from it. Test algorithm for converting MIMNS data into position information and storing it in the memory. Stage 2. Implement obstacle avoidance method explained in the Flowchart 1. below. The unit should be able to move around without crashing into anything, collect information about the path it has taken and store it in the memory. Stage 3. Give unit additional intelligence by implementing a more complex algorithm (not discussed in this report), which will allow the unit to use position information to orient itself and make decision for the next move. It will not aimlessly drive around, but rather be aware of its coordinates at any given time, therefore, be able to make a better judgment which area is still unknown and which part of it has already been covered. Stage 4. In case the three initial stages are completed in time, the attempt will be made to implement a slightly different algorithm which will ensure that the robot has covered the entire area before returning to the initial point. Navigation method which will make this possible is still under investigation by team members. 7. Proposed Work/Schedule 7.1.Milestones Week 3: Unit fully assembled and mobile, with motion control by the microcontroller Week 5: Unit capable of avoiding obstacles Week 6: Unit capable of navigating its way around the loop on the second floor of engineering building

44 44 Week 9(optional): Autonomous unit capable of navigating entire floor of engineering (or any other) building 7.2.Schedule Week 1 Week 2 Week 3,4 Week 5 Purchase all components if not done already Start putting the mechanical part of the unit together Research Microcontroller Operation and Programming Test H-bridge and D.C. motors Keep track and record of work done Proceed constructing mechanical part of the robot Research Microcontroller Operation and Programming Test ultrasonic sensors Test ADC Test D.C motor control trough the H-bridge by the microcontroller Keep track and record of work done Research Microcontroller Operation and Programming Finish up constructing of the mechanical part of the robot Test it Make estimates about the turning or slipping error Implement obstacle avoidance, collect path data Keep track and record of work done Test the unit Implement an algorithm which will utilize MIMNS data in decision making process, make the robot smarter If needed, change sensor location or add additional sensors if found necessary Keep track and record of work done Week 6,7,8 Week 9,10 OPTIONAL (Not in scope) Implement a more complex navigation algorithm which will ensure that the robot covers entire second floor of the building Add additional features in the code which will allow faster sensor response and more accurate readings of the position data If needed, change sensor location or add additional sensors if found necessary Keep track and record of work done Start working on the final report Finish up testing and modifications to the robot

45 45 9 Fabrication and Assembly 9.1 Chassis The chassis is constructed of a combination of off-the-shelve components as well as parts that were custom made for this project. The custom parts were made at RIT s machine shop and off campus at Genesee Group/NY. Team members machined the parts made at RIT with the assistance of Dave Hathaway and Steve Kosciol. While team members machined the parts at Genesee Group/NY with the assistance of Steve Ost, Trevor Young and Bill Bagne. In the design stage, all custom parts were designed to be within the machining capabilities of the team members, so as to eliminate the need for a costly outsourcing of labor. The majority of the parts were made at RIT due to the availability of the materials and tools that were used such as a mill, drill press, lathe, or a combination of these.

46 46 Tolerances of +/-.005 inches were easily met and requirements for flatness, parallelism, and perpendicularity were achieved with good machining practices. The RIT machine shop was used to make the drive shafts (2), the wheel shafts (4), the drive shaft brackets (2), the wheel shaft brackets (4), the sensor fixture (1), the MIMNS brackets (2), and the removable electronics base (1). However, the team members also had contacts with local industries from previous co-op experience that were willing to lend their facilities to our project. We used the facilities at Genesee Group/NY for the chassis (1), motor brackets (2), and servo bracket (1) that would have otherwise taken longer to machine at RIT with a mill. Specifically, their laser resonator and bending machine enabled less time consuming machining than if it had been machined at RIT, allowing us to focus our time on other aspects of the project. The facilities at Genesee Group/NY were used to make the chassis (1), servo fixture (1) and the motor brackets (2). 10 Testing As expected, some unforeseen obstacles arose during the actual assembly and testing of the robot. The Texas Instruments MSP430 microcontroller was more complicated than originally anticipated. The MSP430 manual was not clear in many cases and with no extensive prior experience with microcontrollers, the learning curve was longer than originally anticipated. Several contradictions were found in the manual as well. When understanding the complexity of the MSP430, it was decided to add a second microcontroller to the design (DIOS), and divide the purposes of the two controllers as following. MSP430 was programmed to control the obstacle avoidance and navigation aspects of the project, thus controlling the ultrasonic sensor, infrared sensors, H-Bridges, and DC motors. The DIOS microcontroller was programmed to control the data gathered from the MIMNS and process it.

47 47 Figure 1: Miniature Inertial Measurement Navigation System The DIOS microcontroller would take acceleration readings from the MIMNS, convert the data to displacement calculations, write the results to memory, and later recall these calculations and transmit them to a PC. The DIOS microcontroller was chosen based on its ADC capabilities and the fact that it could easily read the recorded data stored on the EEPROM chip and transmit this to a PC. The DIOS was cheap and userfriendly, but had limited processing power as a result. It would not have the capabilities to perform the obstacle avoidance duties of the MSP430 as well as its data acquisition duties. Using both microcontrollers enabled the team to work in parallel on the different tasks which resulted in a more efficient work schedule. One unexpected problem was the limited amount of timers available on the MSP430. Timers are necessary to control the program flow, the H-bridges, one for the ultrasonic sensor, and one for the servo motor. But since there were only three timers available on the MSP430, we had to devise a method where the function of one timer could be changed to switch between the ultrasonic sensor and the servo motor. The timer is set up to take an ultrasonic measurement, then set up to turn the servo to the desired position to get an ultrasonic reading from, and then reverted to get a ultrasonic measurement reading. The speed of this process was more than sufficient to ensure effective obstacle avoidance at the final speed of the robot.

48 48 During the development stage, the original ultrasonic sensor was damaged beyond repair in a head-on collision with a wall that came out of nowhere. A slightly more expensive ultrasonic sensor was used to replace it (Devantech SRF04); based on better range and durability. However, incorporating the different ultrasonic sensor required some revisions to the code that cost more time. Conclusion and Recommendation While we remain confident that the TI MSP430 could perform all the required functions of both microcontrollers, we simply could not do so within our time constraints. If at all possible, we would have chosen a microcontroller that was more user-friendly but still had all the capabilities of the MSP430. A microcontroller with the same processing power, but with more timers would be ideal. For more time and money, we could have invested in a three ultrasonic sensor system for the front of the robot to achieve the same range as the sensor/servo method. This would have been slightly faster than the sensor/servo method and would have consumed less power, but the control and timer handling software would have had to e rewritten. However, since power concerns were addressed with our more than capable power supply, this was never a problem. Still, it would be a faster method and consume less energy if an attempt to scale down the size and weight of the project was made, allowing the use of smaller batteries. A set of motor encoders would have been useful, but were not worth the money (about $80 per motor). They would have been helpful in evaluating the accuracy of the MIMNS cube by having an alternate measurement of velocity or displacement. A cheaper encoder system could have been made by hand, but time was more valuable in this project than money, and was not essential to the success of this project. Lastly, but most importantly, we considered retrofitting the MIMNS with more sensitive accelerometers. The MIMNS was designed to be used in applications having accelerations of up to 10G s. Meanwhile, our robot rarely experienced accelerations over 2G s. Thus, a large portion of the accelerometers range was never used while measurements in the most critical range were not as accurate as could have been possible, had more sensitive accelerometers been used. These inaccuracies lead to inaccuracies in the path recorded by the robot. While the true shape of the path was always retained, its

49 49 scale typically had a margin of error of about +/- 2 m. This is why the path recorded sometimes shows the robot traveling somewhere where a wall actually exists. More sensitive accelerometers would eliminate this path drift and provide the most accurate path. Figure 2: MIMNSY Figure 3: Generated map from the collected data points from MIMNS Y Displacement (m) Recorded Path Actual Hallway X Displacement (m)