Abstract:
The motion planning of multiple mobile robots undertaking
individual tasks in the same environment is studied. A motion
planning method based on neural networks is proposed. By storing
fuzzy rules in neural networks the method can fully make use of
the association ability and high processing speed of neural
networks to make robots avoid collisions with other objects in
real time. Compared with rules method, the method can not only
avoid building a large and complex rules base but also make
robots avoid collisions and conflicts at higher speed and with
higher intelligence.
Key words:
Multiple
mobile robots Motion planning Fuzzy rules Neural networks
*
This project
is supported by 863 program of China (993502). Manuscript received on
October 9, 1999; revised manuscript April 15, 2000 |