Online collision-free trajectory planning for cylindrical robots with visual feedback
Abstract
When robots work in a partially known or unknown environment, the robot motion based on a trajectory generated off-line in advance may lead to collisions with moving obstacles. On-line testing of possible interferences between the robots and obstacles are necessary. If possible collisions are detected, trajectory modifications are needed for avoiding the collisions. An on-line algorithm is proposed here to test interferences and generate collision-free trajectories for loosely coordinated cylindrical robots operating in an environment with moving obstacles. A method that uses computer vision to detect the position of the moving obstacle in the three-dimensional robot workspace is described. The motion of the obstacle with unknown dynamics is predicted by means of a recursive autoregressive (AR) time series model. The parameters in the AR-model are estimated in the least mean squared-error sense. The maximum velocity and acceleration of the robots are used as motion constraints in the trajectory planning. A control architecture is designed so that all robots in the system generate their own trajectories on-line simultaneously by parallel computations. The approach is demonstrated by simulations. Laboratory experiments for collision avoidance by a Stanford robot and a vision system are described.
Degree
Ph.D.
Advisors
Koivo, Purdue University.
Subject Area
Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.