Evolution of the Autonomous Surface Craft AutoCat

Transcription

1 Evolution of the Autonomous Surface Craft AutoCat Justin E. Manley ', Aaron Marsh2, Whitney Comforth3, Colette Wiseman3 Massachusetts Institute of Technology, 'Research Engineer, Sea Grant College Program 2Currently with Bluefin Robotics, Cambridge, MA 3Undergraduate Students, Department of Ocean Engineering Cambridge, MA Abstract - At the MIT Sea Grant College Program, Autonomous Surface Craft (ASCs) have been under development since These systems have been designed for various missions and have demonstrated success in three separate iterations. The current goal of ASC research at MIT is to develop an easily deployed system that can serve as a tool for educational use, as a precision survey platform, and as a communications and navigation link to an AUV. In the past, this goal has been hampered by limitations in the mechanical design of the current system. Since September 1998 significant improvements have been made to the design of the current ASC. New power, propulsion, and control systems have been installed. These improve the performance of the vehicle and significantly reduce the mechanical and electronic failures that troubled the previous system. In addition, structural modifications have been made to make deployments and operations easier. These design improvements promise to make the ASC a more useful tool for, and provide greater service to, the Autonomous Underwater Vehicles Laboratory at MIT. This paper will briefly present the history of ASC development at MIT. It will then document the design changes made to the current vehicle. The problems experienced with the old designs will be identified and the improvements represented by the new systems will be explained. The paper will conclude with a brief discussion of the next mission for the ASC. A. The ASC ARTEMIS I - ASCs at MIT Sea Grant The first ASC produced at MIT Sea Grant was named ARTEMIS. This vessel is a 1/17 scale replica of a fishing trawler (with a total length of 137 cm) that was originally used for model basin testing. Installation of an electric motor and a servo actuated rudder made the basic model into a platform capable of testing the navigation and control systems required by an ASC. Initial work focused on the development of control systems for the ASC. A microprocessor and digital compass were installed to provide rudimentary navigation and control functions. This configuration used a proportional-plus- derivative (PD) control system to implement simple heading control. These first steps yielded an ASC with limited autonomy but provided a valuable proof of concept [ 11. Follow on work added a Differential GPS (DGPS) receiver to enhance the navigation system of ARTEMIS. This facilitated the development of a heading constrained waypoint following controller based on Fuzzy Logic. A depth sounder was added and ARTEMIS executed waypoint-defined surveys to generate bathymetric maps of the Charles River in Cambridge, MA. The addition of a radio modem allowed these bathymetric maps to be generated in real time and provided human supervisory control of the ASC [2]. Upon completion of these experiments ARTEMIS was retired from active use. B. The Autonomous Kayak One of the primary shortcomings of the ASC ARTEMIS was its small size. This limited its endurance and seakeeping. The field operations of ARTEMIS were limited to the Charles River, a region of limited scientific interest. To produce an ASC with more useful capabilities a kayak platform was examined [3]. Upon completion of design studies a 3 m long kayak hull was obtained and converted into an ASC. A new propulsion system and actuator were installed, as were electronics similar to those used in ARTEMIS. This new vehicle underwent a series of trials on the Charles River. It was then fitted with acoustic tracking systems and used to follow a tagged fish [4]. Upon conclusion of these experiments this ASC was turned over to the MIT Ocean Engineering Teaching Lab for use in student research and education projects. C. The ASC ACES To continue the automated bathymetry experiments begun with ARTEMIS, a new ASC was developed. The specifications of the next ASC were based on a desire to create a system as versatile and useful as a small manned vessel while maintaining a small size to allow for easy deployment and survey operations. As ARETMIS was slow, unstable, had poor endurance and a small payload the next ASC represented a significant improvement in platform capabilities. The new ASC ACES (for Autonomous Coastal Exploration System) was developed during 1996 and 1997 [5] /00/$ IEEE 403

2 The mechanical design will be discussed further below. The electronics suite and control software were directly ported from the ASC ARTEMIS and incrementally improved over the development of the system. The completed ASC underwent field tests off Gloucester, MA during the summer of Upon completion of these trials it was outfitted with sensors suitable for hydrographic survey and successfully completed such a survey in Boston Harbor in December 1997 [6]. Beginning in January 1998 the ASC ACES was returned to the lab for a significant upgrade of its electronic systems and software. This effort was not completed due to a change in the software strategy at the MIT AUV Lab. It was decided to postpone improvements to the ACES electronics and software until a new AUV design was completed and tested. This work was recently completed and the software and computer systems standard developed for the new AUV will be modified and applied to the ASC in the near future. Between September 1998 and June 2000, while the AUV systems were developed, the mechanical systems of the ASC were heavily modified. During the extensive field-testing of ACES several potential improvements in the basic platform design were identified as desirable. These improvements have been made and are the subject of the remainder of this paper. I1 - Mechanical Systems, Original Configuration The original configuration of the ASC is presented in Fig. 1 at the end of this document. A. Hull and Structure The platform of the ASC ACES is based on a commercially available recreational vessel, the Hobie Float CatTM. The Float Cat's rotationally molded polyethylene hulls (1.8 m long) come with a light metal structure and web seat, which was replaced by a custom frame. The ASC frame consists of four stainless-steel longitudinal members and four more stainless-steel transverse members. These members are joined together by PVC mounting blocks. The inner two transverse members are tubes of the same diameter as the Float Cat's standard cross pieces. This allows the hulls to slide over the cross bars and be secured with quick release pins. The overall beam of the ASC using this frame is 1.3 m. The steel frame was covered by a polyethylene deck for mounting of light subsystems like the GPS antenna. More integral systems including the rudder and motor mounts were bolted to the frame. A thin plate with threaded holes was inserted inside the longitudinal cross pieces, which were then drilled with matching holes. This provided a very secure mounting system with a common bolt pattern. This design was chosen for its high strength and corrosion resistance. It was envisioned that the quick release mounting of the hulls would provide for easy removal of the hulls and thereby improved transportation of the ASC on land. Additionally, locating the mounting plates internal to the longitudinal members was expected to provide a simple mounting interface for future needs. These were worthy goals but the actual design was only moderately successful. During the extensive testing of the ASC it was discovered that the structural frame was actually quite a hindrance to field operations. The first complaint was weight, the frame alone weighed over 20 kgs. The main vehicle computer, rudder servo, and electronic supply batteries were all mounted to the frame adding to the weight of this component. This represented the majority of the ASC and it was not practical to remove the components to reduce the weight. This was counter to the original expectation that the vehicle could be partially disassembled and transported modularly. While the motor was always removed for transport to a test, it was discovered that, due to excessively fine tolerances in the design of the main transverse tubes, and despite the quick release mechanism, the hulls could not be easily removed. Even if they were removed, the main transverse members still extended out to the complete 1.3 m beam of the ASC. While this width was just within the constraints of a small pickup truck, allowing ACES to be deployed in the field with relative ease, it did not allow for easy transport inside labs and test tanks. A hrther problem with the modularity concept was the mounting system for components. While the internal mounting plates were effective they were not especially easy to work with, as they required a matching bolt pattern on any major component. The housings for the electronic supply batteries and the vehicle computer did not have this matching pattern. Therefore, they were mounted directly to the deck with through-bolts and could not be quickly removed for transport. In summary, the originally structure of the ASC ACES was quite strong and corrosion resistant but it did not fulfill the planned requirement of modularity. This led to the ASC being excessively heavy and overly wide for easy deployment in field and tank tests. B. Power and Propulsion The original power and propulsion systems for ACES provided exceptional performance. A small gasoline outboard motor was used for propulsion. This provided 2 hours of operation with just the internal tank on the motor. The separate batteries used to power the electronics could provide over 8 hours of vehicle run time. This was deemed suitable for experimental work and there was sufficient payload to increase fuel and battery capacity so the ASC could run for up to 12 hours. The gas propulsion also provided high speed. While accurate measurements of the ASC's top speed were not made, it approached an estimated 20 knots during radio-controlled tests. This potent propulsion system was impressive, albeit highly complicated. Operations required the motor to be started by an operator pulling on the hand crank and then shifting the motor into forward gear when the mission began. From that point the ASC computer used a custom built actuator to control the throttle. The ASC computer would shut the motor down at thc 404

3 end of a mission by closing the throttle. At this point, operator intervention was required to retrieve andor restart the ASC. Original plans called for additional custom actuators to provide the starting and gear shifting functions. During testing it was discovered that maintaining the custom throttle system in the field was complicated. In the event of software failures the computer frequently "forgot" the throttle setting. As there was no manual adjustment for the throttle, a complicated sequence of sohare steps was needed to set the throttle for starting. Without adding a feedback sensor to the throttle actuator it would have been quite difficult to program the ASC to restart its e 1 f. Given the challenges of implementing just one computer actuator for the gasoline outboard motor, it was decided that developing a reliable starter and gear actuator would have required an inordinate amount of engineering effort. This reduced all field operations to the operator intensive process described above. This was acceptable for trials and basic tests but was not appropriate for wide scale use of the ASC. Despite its complications, the performance of the ASC power and propulsion systems was excellent. High speed and long range are desirable traits in an ASC, as is the use of internal combustion with its excellent energy density advantages. Unfortunately, for an automated system the gasoline outboard motor was too complex for reliable operation. C. Steering Actuator Steering systems on the first generation of ACES provided good performance but only adequate reliability. The actuator consisted of a stepper motor turning a surfboard skeg through a series of gears. The entire system was housed in a custom built watertight enclosure. The turning performance of the ASC using this system was quite good. In radio-controlled tests it performed sharp turns at high speeds and could complete a turn in just over the length of the ASC. Unfortunately, reliability issues plagued this system as well. There were two weak points in the rudder design. First, strong blows to the rudder could force the gears to turn inside the housing. This could loosen all of the fasteners and required the entire system to be removed, disassembled and then reassembled to ensure mechanical integrity. While undesirable this event was rare and usually occurred during movement of the ASC in the lab. Removing the rudder from the mounting point satisfactorily reduced the vulnerability to this failure. The second weakness in the original design of the rudder system was the lack of a feedback sensor. During trials the rudder was set to the zero position manually and the computer tracked changes in its position during missions. In the event of software or computer failures the rudder position would be lost and it would need to be set to zero again manually. The lack of a feedback sensor was attributable to short timelines and limited resources. Long-term plans always assumed that such a system would be added Mechanical Systems, Updates and Improvements The modified configurations of the ASC are presented in Fig. 2 and Fig. 3 at the end of this document. A. Hull and Structure The hull of ACES had proven to be a wise choice as it was inexpensive, light, and performed quite well. Therefore it was decided that the basic catamaran hullform using the commercially available Float CatTM hulls would be maintained. The heavy stainless steel structure needed to be modified and eventually replaced. The first stage of modifications addressed the lack of modularity in the original design. The main cross tubes were removed and cut in half. They were then slightly shortened. The tubes halves were then permanently mounted to the hulls. To mount the hulls the original quick release pins were retained. When these pins were removed the hulls and cross tubes could be removed reducing the width of the deck structure to 63.5 cm. This was a much more manageable size for modular transport of the ASC. To reduce the weight of the original ASC deck module the batteries were removed from the watertight housings secured to the main deck. Compartments were built within the hulls to house the batteries. The fore and aft sections of each hull were filled with foam and the midsection was reinforced with plywood bulkheads. The compartment between the bulkheads was left open for the batteries and a watertight lid was installed over the opening in the hull. This change reduced the weight of the deck module and increased the stability of the ASC by moving the center of gravity of the heavy batteries below the waterline. This structural configuration was tested in the Charles River during July 1999 and demonstrated performance equal to the earlier configuration. Given the improved modularity of the ASC this was considered a success. However, the complete ASC was still quite heavy. To reduce the weight of the structure the stainless steel frame and polyethylene deck were eliminated. A new lighter structure was built. ExtrenTM (fiberglass plastic composite) angles were assembled into a rectangular frame, 122 cm long by 61cm wide. The width of the frame was slightly reduced from the original configuration to improve mobility ashore. The comers of the frame that secured the angles were made of solid PVC. This open rectangle was then covered with a PVC sheet cm thick. This new integrated deck and frame weighed slightly more than 10 kg. This new structure was less rigid than the stainless steel version but installation of a lumber "transom" amidships stiffened it sufficiently for operations. Currently this new structural configuration has only been tested in a tank setting but the design improvements have greatly simplified deployment of the ASC in a tank. The reduced weight and increased modularity of the new ASC structure adequately address the flaws of the original design. 405

4 B. Power and Propulsion Substantial changes were made to the power and propulsion system of the ASC. The complicated outboard motor was eliminated in favor of a MinnKotaTM electric trolling motor. Initially, the basic configuration of the vehicle was maintained. The new electric motor was installed in the same location as the outboard. To support the larger power requirements the small batteries that originally powered just the electronics were replaced with two heavy-duty 105 amp hour gel cells. These batteries were housed in new compartments as described above. This single electric motor configuration was tested in the Charles River in July Performance was not as impressive as the outboard configuration but still excellent. Speeds up to 7.3 knots were observed and the larger batteries were calculated to provide at least 6 hours endurance. The one flaw in this new configuration was the continued lack of reverse power. The motor controller selected for the trolling motor was capable of running at 75 amps, safely above the maximum 27 amp draw of the motor, but it did not offer a reverse mode. To overcome this problem it was decided to replace the single motor with twin electric motors. A second trolling motor identical to the fvst was procured. The transition to twin screws was made at the same time as the installation of the new structure described above. Both motors were mounted amidships on a lumber "transom" just inside the hulls. Using two motors allowed the maximum current draw for propulsion to be cut in half, to roughly 15 amps per motor. Fuses and software settings maintain this limit. This current load reduction meant that a different motor controller, with a current rating of 25 amps, could be installed. This new motor controller was capable of driving the motors in reverse. With both configurations, AstroflightTM motor controllers were used to drive the motors. Pulse-widthmodulation signals were sent to the controller from a TattletaleTM TT8 microprocessor. Initially, the controller was housed in a PVC cylinder mounted to the motor and the TT8 was contained in the rudder system housing. When the configuration was changed to twin motors the TT8 and both motor controllers were housed in the main electronics box. Switching to electric motors proved to be an excellent choice. Controlling the motors, including throttle adjustments and stopping and starting, is much simpler. The TT8 system is much easier to interface than the custom mechanical throttle actuator. Manual intervention is not required at any point as all functions are now fully electronic. This new propulsion configuration provides for enhanced maneuverability, notably the ability to reverse the thrust, and significantly improved reliability through the removal of all moving parts with the exception of the propeller. Performance, both speed and endurance is comparable to the previous propulsion system. C. Actuators As noted above, the primary flaw in the original rudder system was the lack of a feedback sensor to track absolute rudder position. During the academic year a student at MIT designed a new rudder controller for the ASC. This design included the use of an optical encoder for rudder feedback sensing and a TT8 microprocessor for overall control of the rudder system [7]. This design was implemented and tested in the Charles River during the July 1999 tests. The feedback sensor worked well and the TT8 provided effective control of the rudder. It should be noted that this TT8 also provided the motor control signals. While the new rudder system overcame the software and control problems of the original design, it was still vulnerable to mechanical failures. The rudder system exhibited a "stickiness" during the river trials. Even with enhanced feedback and control the mechanical complexity of the rudder system was a weakness. It was decided that eliminating the rudder mechanism was the best option to improve reliability. With the installation of two motors, both capable of reverse thrust, it is now possible to steer the ASC entirely by thruster actions. In fact, this provides for even better turning performance. Using the rudder system the ASC had a turning radius roughly 1.5 times its length. The new configuration allows the ASC to turn on its own center axis. The use of twin screws to maneuver the ASC was tested in a tank at the MIT Department of Ocean Engineering. As with the new deck/frame structure this third configuration of the ASC demonstrates significant improvements. The ASC turns quite well and its straight line tracking is still excellent. For typical "lawn mower" survey patterns the ASC has achieved nearly optimal maneuverability. IV Conclusions and Future Work The series of mechanical improvements made to the ASC since June 1998 all proved quite successful. The primary sacrifice has been a reduction in top speed from over 20 knots to around 7 knots. Since all scientific missions envisioned for the ASC require slow cruising this is an acceptable change. All other changes in the ASC performance were significant improvements. Operationally the vehicle is much simpler. No manual intervention is required to begin, end, or resume missions, nor are there servo system to be calibrated at each boot up of the vehicle computer. These changes are most welcome. The enhanced reliability and maneuverability of the ASC have also improved with the new designs. In honor of these improvements, and as a thank you to the Hobie Cat Company whose hulls have been the only consistently used part of the design, the ASC has been renamed AutoCat. The next step in ASC research at MIT Sea Grant is application of the new vehicle to a current research project. Autocat will be used to perform a high precision sub-bottom profiler survey of a shipwreck. A closed-loop control system using precision acoustic tracking is under development. Thiz will allow the ASC to navigate with 3-5 cm resolution. Thc improved tracking and maneuverability, combined with thicontrol system, will yield a very precise survey of the wreck 406

5 site. This data set is of great interest as is the systems integration challenge of collecting the data. The successful completion of this survey will fully validate the design improvements described here. Acknowledgments The authors would like to acknowledge the support of their colleagues at the MIT Sea Grant College Program. The participation of MIT students was made possible by the MIT Undergraduate Research Opportunities Program and is greatly appreciated. The MIT Sea Grant College Program, under grant number NA86RG0074, supported the work described here. References [ 11 J. Manley and M. Frey, Development and Operation of the Autonomous Surface Craft ARTEMIS. MIT Sea Grant Undergraduate Summer Research Program [2] T. Vaneck, J. Manley, C. Rodriguez, and M. Schmidt, Automated Bathymetry using an Autonomous Surface Craft, NAVIGATION, Journal of the Institute of Navigation, [3] J. Manley, A Preliminary Design Study for an Autonomous Surface Craft Society of Naval Architects and Marine Engineers New England Section Student Paper, February, [4] C. Goudey, T. Consi, J. Manley, M. Graham, B. Donovan, L Kiley, A Robotic Boat for Autonomous Fish Tracking, Marine Technology Society Journal Vol. 32 No. 1, Spring [5] J. Manley, Development of the Autonomous Surface Craft ACES, Oceans 97 MTSAEEE Conference Proceedings, October, [6] J. Manley, T. Vaneck High Fidelity Hydrographic Surveys Using an Autonomous Surface Craft, Oceans Community Conference 98 MTS Conference Proceedings, November, [7] M. Schmidt, Electromechanical Design of a High Performance Rudder Controller for an Autonomous Boat, Senior Thesis, MIT Department of Mechanical Engineering, June,1998 Fig. 1: The ASC ACES configured for hydrographic survey, June

6 Fig. 3: The ASC AutoCuf in its most recent configuration, June