Trajectory Planning and Control of Industrial Robot Manipulators

摘要

Industrial robot manipulators are used in various applications in order to achieve fast, precise, and quality production. In pick-and-place operations such as part handling, assembly, etc., the end-effector of the manipulator has to travel between two specific points in the workspace, and the path it takes in between is of no concern. In trajectory tracking applications such as welding, cutting, painting, etc., the end-effector has to follow a specific trajectory in 3space as closely as possible, while maintaining rated velocity as much as possible (Munasinghe, 2001). In the latter case, planning the trajectory can be complex when there are constraints on the end-effector velocity, joint acceleration, and trajectory error. Trajectories planned without proper consideration to these constraints often result in poor performance such as trajectory overshoots, end-effector deviations from the planned trajectory, and undue velocity fluctuations (Nakamura, et. al., 2000). Performance could be even more deteriorated especially at sharp corners in the Cartesian trajectory (Nakamura, 20001). Lot of trajectory planning algorithms have been proposed so far starting from simple Cartesian path control (Paul, 1979) to time optimized trajectories (Shin, 1985). However, the industrial systems experience difficulties accommodating most of these methods because of at least two specific reasons; 1) These techniques often require hardware changes in the existing setup and the manufacturing process has to be interrupted for system reconfigurations, which usually takes a longer period of time, and 2) Many of these methods often consider only one constraint, and often they pay less concern about industrial requirements and actual constraints set by applications. Therefore, they find difficulties in industrial implementation. In this view, we present a new trajectory planning algorithm which considers end-effector velocity limit, joint acceleration limit, and error tolerance set by the application. These are the actual constraints in most industrial applications. Another technical problem in industrial manipulators is their delay dynamics, which causes the end-effector to overshoot at trajectory corners. To remedy this problem, we have designed a feed-forward compensator (Goto,