COLLISION-FREE PATH PLANNING ALGORITHM FOR TWO-LINK ROBOTIC ARM BASED ON GEOMETRIC MODELING

The evolution of learning based intelligent path planning in robot manipulators requires beforehand controllable amount of data to conduct its training and testing activities. In this paper, a systematic collision avoidance path planning algorithm is developed to offer these training data. The importance of the developed algorithm is the reality that generalizing a solution from training data depends largely on the data representation scheme. Irrespective the initial positions of the robot links, the proposed algorithm detects the goal and obstacle positions using infra-red. detectors, determines the reachable work space area, recognizes all possible paths to the goal, checks the obstacle position relative to the recognized paths, then specify the most suitable path that avoids collision with obstacle. The simulation of the proposed algorithm proves its ability to generate any pre-determined number of records random initial links positions, random obstacle positions, random goal positions, and their corresponding final adjustments in links angles required to cover the reachable work space area with collision avoidance. Simulation of the proposed algorithm proves its higher flexibility in controlling the number of resulted records which consequently controls the accuracy of any learned based intelligent path planner intended to be implemented in a future research work.