Obstacle avoidance algorithm based on biological patterns for anthropomorphic robot manipulator

This study addresses the problem of collision-free controlling of 3-DOF (degree of freedom) anthropomorphic manipulators with given a priori unrestricted trajectory. The robot constraints resulting from the physical robot's actuators are also taken into account during the robot movement. Obstacle avoidance algorithm is based on penalty function, which is minimized when collision is predicted. Mathematical construction of penalty function and minimization process allows modeling of variety behaviors of robot elusion moves. Implementation of artificial neural network (ANN) inside the control process gives the additional flexibility needed to remember most important robot behaviors based on biological pattern of human arm moves. Thanks to the fast collisions' detection, the presented algorithm appears to be applicable to the industrial real-time implementations. Numerical simulations of the anthropomorphic manipulator operating in three dimensional space with obstacles is also presented