WORK PARTNER - HUT-AUTOMATION S NEW HYBRID WALKING MACHINE

Transcription

1 WORK PARTNER - HUT-AUTOMATION S NEW HYBRID WALKING MACHINE Ilkka Leppänen, Sami Salmi and Aarne Halme Automation Technology Laboratory Helsinki University of Technology PL 3000, HUT, Finland ilkka.leppanen@hut.fi ABSTRACT. The paper introduces a new platform with hybrid locomotion capability. It is designed for a service robot, called WorkPartner, used mainly in outdoor environment when doing work interactively with human operator. The estimated weight of the platform is about 160 kg and the pay load about 60 kg. The actuation system is fully an electrical one and the power system a hybrid one with battery and a 3 kw combustion engine. The locomotion system includes legs with foot replaced by a wheel. It allows motion in three mode, with legs only, with legs and wheels powered at the same time or with wheels only. With wheels the machine is designed obtain 12 km/hour speed on a hard ground. The purpose of the hybrid locomotion system is to provide a rough terrain capability and a wide speed range for the machine at the same time. The paper describes the basic design of the new machine and introduces some theoretical analysis of the hybrid locomotion mode. the power system a hybrid one with batteries and a 3 kw combustion engine. The locomotion system allows motion with legs only, with legs and wheels powered at the same time or with wheels only. With wheels the machine can obtain 12 km/hour speed on a hard ground. The purpose of the hybrid locomotion system is to provide a rough terrain capability and a wide speed range for the machine at the same time. The mechanical design, mainly reported in this paper, has been made in cooperation with Russian Rover Company Ltd (St Petersburg) specialised on space robotics. The company is also presently manufacturing the prototype platform. The WorkPartner 1, the first generation prototype, will be ready at the end of 2000 and the third generation prototype WorkPartner 3 at the end of Introduction. After the successful ten years research with the MECANT - machine [2] the Automation Technology Laboratory at HUT has started the project to construct a new generation walking machine which will be a prototype service robot for multi-tasking work. We call the robot WorkPartner, because the idea is to make a highly adaptive robot, which can carry different tools and work interactively with a human person by learning at the same time the details of the task. The platform, shown in Fig. 1, on which the robot will be built is called HYBTOR (Hybrid Tractor). It has four legs equipped with wheels and an active body joint. The weight is estimated to be about 160 kg and the pay load about 60 kg. The actuation system is fully an electrical one and Figure 1. The HYBTOR platform. 2. Requirements for the design of the WorkPartner Future service robot must be able to work interactively with people and simultaneously learn along the tasks. Such properties requires a lot from the WorkPartner; superior mobility, learning capability, a user interface which allows communication on the level of cognition, and an autonomous operation

2 capability as navigation and obstacle avoidance in the working environment. Typical, and reference tasks in mind are for example light forest and park-area harvesting works ( pre harvesting, gardening ), transport and cleaning tasks in industrial environments or urban areas and guarding. Requirements for a platform Because the WorkPartner works with people, it must have almost the same mobility as people do. This means mobility on uneven, outdoor terrain, capability of moving in sand, swamps, snow and rocky grounds. In indoor environments it means capability of moving in staircases upwards and downwards. It also must be able to work indoor and/or outdoor surroundings. The robot must have wide speed range to be able to move along running or cycling man. The weight should be about the same as a human person has. The energy consumption must be low to guarantee a long operation time, say several hours of work. The machine must be environment friendly, meaning gentleness when moving and low emission. The payload capacity must be enough to carry along light manipulators, tools and the possible articles needed in the work. The sensing and navigation system of the robot should allow flexible perception of the environment and navigation in it. In addition transferring cognition level information to the user trough a virtual word description common to both the robot and its user. At this stage of the project there is only a pale presentiment about the sensor technology needed, but it probably will be based on fused vision, lasar, US and possibly RF-sensor information. Also inertia and force sensors installed in the body will be used effectively. 1.2m. The weight is about 160kg and payload about 60kg. The actuation system is fully an electrical one. Brustless EC-servo motors (Maxon) are used. The power system is a hybrid 48 volt system with batteries, a generator and a 3 kw combustion engine. The batteries provide the energy for high power peaks and the generator is used to reload the batteries. The power from the generator is quite enough for transition drive with wheels. The locomotion system consists 4 wheeled legs and an articulated body. Each of the leg structures are similar, the leg itself is 3 degree of freedom mammal type leg, the foot being replaced with an active wheel (motor inside the wheel). The control system hardware consists of four micro-controllers (Siemens, one for each leg), an industrial pc104 Pentium processor as the main computer, and a CAN-bus which connects all of them. The micro-controllers control the drive motors trough a drive electronics specially designed for EC-servo motors. The software is being developed under QNX real-time operating system. The wheeled leg: The leg joints linear screw actuators are driven with 250W EC motors and the same motor is used in the wheel, too. The speed of a motor is measured with a hall sensor, and the positions of the joints are measured with potentiometers. The force directed to the wheel is measured with special made force sensors installed inside each foot. The working volume of a leg can be seen in Fig. 2. In this paper we concentrate on the mechanical design of the HYBTOR-platform and will report about the next steps later on. 3. HYBTOR -platform Description of the platform: The physical measures of the platform are length 1.40m, width 1.20m and height 0.5-

3 fan effect of the rotating tire on the surrounding air also contribute to the rolling resistance of the tire, but they are of secondary importance. Figure 2. Working area of a wheeled leg. The payload of the machine is planned to be 60 kg in the walking mode, but with leg joint actuators locked and only wheels active, the platform can carry a payload which is much more. 4. Locomotion modes Propulsion system consisting of combination of leg and wheel provides powerful locomotion capability. The machine has three locomotion modes: walking locomotion mode wheeled locomotion mode hybrid locomotion mode In the walking mode the wheels are locked and locomotion utilises the legs only. This locomotion mode fits for uneven terrain when moving speed is low (below 5 km/h). To reduce energy consumption conventional wheeled locomotion are used in higher speed (5-15km/h). Hybrid locomotion is combination of wheeled and walking locomotion. Wheeled locomotion mode Wheeled machines, like cars and trains, have a superior payload capacity and speed on hard ground. The rolling resistance of tires on hard surfaces is primarily caused by the hysteresis on tire materials due to the deflection of the carcass while rolling. Friction between the tire and the road caused by sliding, the resistance due to air circulating inside the tire, and the However, the wheel does not work well in soft ground, in off-road operation. When the tire sinkage is significant, a bulldozing resistance also is taken into account in the calculation of the total motion resistance of a tire. Motion resistance, which depends on ground deformation, is very hard to measure reliably; using energy consumption or wheel moment information the motion resistance can be estimated. To reduce rolling resistance in offroad operation a diameter and width of a tire will be increased [5]. The lower tire inflation pressure decreases slip of a tire and reduces ground penetration which means the minor motion resistance, too [3]. Quite often the rear tires of a vehicle travel in the ruts formed by the front tires and the motion resistance of the rear tires is smaller. However, on deformable surfaces, such as sand or snow, the motion resistance of a tire can be too big to locomative with wheels. The average values of coefficient of rolling resistance fr for various types of tires over different surfaces are summarised in figure 3. TIRE TYPE SURFACE Concrete Medium Hard soil Sand Passenger car Truck Tractor Figure 3 Coefficient of rolling resistance [5]. A leg equipped with wheel enables an active suspension, force distribution on each wheel, adaptive ground clearance and body level control. Walking locomotion mode Walking locomotion is suitable for uneven terrain, especially in soft ground, where it is impossible to go with wheels. To choose the walking mode two criteria may be used: energy consumption to avoid environmental damage

4 Comparing measured energy consumption values of walking and wheeled locomotion the right locomotion mode may be chosen even automatically (one of the interesting study objective in the future). Energy consumption of walking does not depend so much on ground properties, like softness, and estimated consumption values can be utilised, too. The other criterion to use legged locomotion is to avoid environmental damage. In that case energy consumption of walking can be even higher. The problem to use environmental damage as a criterion is that there are not the standard ways to measure damages. The choice quite naturally remains to the user of the robot. It is important also to notice that walking enables omnidirectional moving in uneven terrain. Hybrid locomotion mode In a hybrid locomotion mode legs and wheels are used actively at the same time. In this mode the wheel is mostly locked in support phase and the leg works. This is because of the driving wheel pull is much less than the pull produced by the blocked wheel (Bekker, 1969). In the recovering phase the wheel is actively rolling and following unevenness of ground. Because of the tire is contacting the ground all the time the recovering leg can support partly the body which increases stability. During the recovery phase the leg can be switched to support phase rapidly if needed. The inertia of the leg limits the velocity of the walking machine using stabile gaits. The recovery leg must accelerate very fast and then slow down in order to prevent collision. With six-legged machines this can bee seen in fig. 4 below. Figure 4 Transfer foot velocity vs. locomotion speed [4] In hybrid motion the wheel can accelerate the leg in the recovery phase and the acceleration can happen a little bit before than the leg is totally switched on lifting phase. At the end of the recovery phase decelerating can happen much more slowly due to the rolling wheel e.g. the foot speed is not zero in relation to the ground. Switching between the support and recovery phase occurs thus steplessly. The hybrid mode can be used also as a skiing mode when the wheels generate actively speed while walking. This may increase the speed considerably and help moving on very slippery surfaces. 5. Conclusion and future work Our work to construct the new platform with hybrid locomotion capabilities is only starting now. The first leg tests will start in November 1998, the platform should be mechanically ready in April 1999 and the first prototype of WorkPartner-robot functionally ready at the end of At this phase of development we are interested to study the basic locomotion features and development of the motion control system which supports hybrid locomotion. In addition to the advantages of hybrid locomotion described above, there are other properties which are under studying. For example, the possibility how terrain following helps to solve a collision problem of a foot and help to percept mechanically the close vicinity of the robot when it is moving. In off-road conditions it is hard to measure the distance to the ground in order to adjust the foot speed before contacting the ground. The wheeled leg in recovery phase can be used as a sensor to measure unevenness of

5 the ground or the wheeled leg can get down smoothly thanks to the rotating wheel. In general, each leg must provide the certain force to keep the body stabile or to accelerate the body. Due to the different ground properties it is useful, for instance, to let the left side wheeled legs act as wheels and the right side ones as pure legs. So, each leg can work in different locomotion mode e.g. in the wheel, walking or hybrid mode. This may help considerably to move the robot out from a slippery ground or a hard stacked position. References 1. Bekker M. G. (1969). Introduction to terrain-vehicle systems. The University of Michigan Press. U.S.A. 2. Halme A., Hartikainen K. and Kärkkäinen K. (1994). Terrain adaptive motion and free gait of a six-legged walking machine. Control Eng. Practice, Vol. 2 No.2, pp Gillespie T.D. (1992). Fundamentals of vehicle dynamics. Society of Automotive Engineers, Inc. U.S.A. 4. Todd D.J.(1985). Walking machines: an introduction to legged robots. Kogan Page Ltd. 5. Wong J.Y. (1993). Theory of ground vehicles. John Wiley & Sons, Inc. Ottawa, Canada.