Name
Abstract | ||
|---|---|---|
This paper presents a method for learning and generating rhythmic movement patterns based on a simple central oscillator. It can be used to generate cyclic movements for a robot system which has to solve complex tasks. The system is laid out in such a way that multiple motion dimen- sions, or degrees of freedom of the robot, are represented independent of each other; therefore, an extension to higher- dimensional problems is easily possible. Guiding the robot by holding its end-effector, the user teaches simple movement primitives forming the basis for a more complex task. Each movement primitive is represented in the system using an oscillator combined with a learned nonlinear mapping. These primitives are then optimally combined to a complete solution to the posed problem. Said optimality is obtained using simulated annealing with the A global search algorithm. Our approach is demonstrated on the problem of wiping a table, but can be used for many typical problems in service and household robotics. I. I NTRODUCTION While most present-day applications of robots are still restricted to industrial environments, the field of service robotics has a huge potential for growth. The development of lightweight arms and hands as well as the progress in robot navigation systems (mobile platforms) on one side, and the development of advanced humanoid robots on the other, provide the hardware premises for a break-through in this field. Increased sensor feedback capabilities (force- torque, visual, tactile, etc.), and the growth of comput- ing power give the possibility to react more flexibly on changing environments. But providing the robots with an appropriate degree of autonomy which makes them able to use these basic reactive features, as well as providing a simple and efficient human user interface, are still very challenging research topics. Especially the robot program- ming task, done by the—often technically unskilled—user, has to be as simple as possible. In this concept, program- ming by demonstration is a widely accepted paradigm. The usual approach, in which the human demonstrates a complete task and the recorded trajectory is followed by the robot, would certainly lack the flexibility needed to survive in common household environments. These environments have a continuously changing, complex structure. A possible solution is the use of small motion primitives which the user teaches to the robot. Equipped with the appropriate algorithms, the robot uses these primitives in combination with sensory input to accomplish a family of complex tasks. In this paper, we demonstrate such algorithms which may be generally useful in this context. We demonstrate our approach in an application which consists of wiping surfaces which are previously specified by a user or detected by a vision system. Only a "wiping style" has to be taught by the user, in the form of a primitive movement pattern. The solution presented in this paper is focused on the following topics: • teaching and learning phase of the primitive move- ment using a torque-controlled manipulator; • providing a trajectory generator (as an alternative to usual trajectory interpolation), which can combine these primitive movements to a smooth trajectory and which reacts adequately to external disturbances. Such disturbances could be obstacles or humans which may interact with the robot during execution; • developing an algorithm which allows automatic movement generation from the taught primitives; • execution of the entire task in a realistic environment, using the same manipulator, in order to validate the approach and test the required robustness and reactiv- ity. |
Year | DOI | Venue |
|---|---|---|
2004 | 10.1109/IROS.2004.1389957 | IROS 2004). Proceedings. 2004 IEEE/RSJ International Conference |
Keywords | Field | DocType |
end effectors,learning (artificial intelligence),service robots,simulated annealing,A* global search algorithm,autonomous service robotics,cyclic movement generation,end-effector,household robotics,learned nonlinear mapping,repetitive movements,rhythmic movement pattern generation,simple central oscillator,simulated annealing | Robot learning,Simulated annealing,Computer vision,Social robot,Robot control,Search algorithm,Computer science,Control engineering,Robot end effector,Artificial intelligence,Robot,Robotics | Conference |
Volume | ISBN | Citations |
4 | 0-7803-8463-6 | 8 |
PageRank | References | Authors |
1.00 | 8 | 3 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| holger urbanek | 1 | 8 | 1.00 |
| Alin Albu-Schaffer | 2 | 2831 | 262.17 |
| P. Patrick Van Der Smagt | 3 | 274 | 35.19 |