Collision avoidance and control of multiple revolute manipulators
Abstract
In chapter 1, a survey on collision avoidance for multi-robot systems is presented. In chapter 2, the problem of path planning for a revolute two-link arm, operating in a workspace filled with obstacles whose boundaries are enveloped by circles or possibly any other shape, is addressed. The path planning approach presented in this chapter is developed in the manipulator joint space and consists of two strategies. The first strategy is used in tightly packed free spaces located between the workspace obstacles, this approach relies on constrained optimization. The second strategy is used whenever the manipulator is operating in any free space region far from the workspace obstacles, this strategy does not use optimization but relies on a global but safe approximation of the free regions of the joint space. This division of the path planning approach reduces the chances that the manipulator will get stuck in a geometrical local minimum during the planning procedure, compared to what would likely happen if the first strategy is used throughout the entire path planning procedure. With the approach taken here, the chance of finding a path for the manipulator from its given start point to its desired goal point in an automated manner is high in comparison to one which uses just one of the two approaches. In chapter 3, the problem of path planning and control of multiple robots operating in a common workspace is addressed. The developments in this chapter merge the control and trajectory planning issues for obstacle avoidance into one. An on-line control law has been determined which avoids collisions between the arms while guiding the end-effectors from their respective starting positions S to their respective desired goal positions G, if such a collision free trajectory exists. The control law is determined from a Lyapunov function based on the total system energy. Simulation results are presented which show that the control scheme developed in this chapter is able to guide two manipulators working in a common workspace avoid collisions with each other as they reach their goal points. No off-line path planning involving geometric analysis is necessary for this scheme to work if a collision free trajectory is feasible.
Degree
Ph.D.
Advisors
Ahmad, Purdue University.
Subject Area
Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.