OPTIMAL PATH PLANNING FOR ROBOT MANIPULATORS

The study reported in this paper deals with a computer-based methodology for the synthesis of an optimal tool path for robot manipulators in the presence of obstacles and singularities of the workspace. The methodology plans optimal path to achieve the best robot kinematic and dynamic performance criteria formulated through proper objective functions. The algorithm uses robot design parameters, the size and the location of the obstacles, and the initial and the goal states to generate a collision-free optimal tool path. Using these inputs the robot workspace is generated and discretized, and the obstacles are modeled as forbidden regions of the workspace. The search for the optimal path begins with the definition of a searchspace that includes the starting and the end points. All possible paths in the search space connecting these points are enumerated through the formation of a network graph structure. An intelligent heuristic search scheme has been developed to enumerate the network of allowable paths. The optimal path is then obtained as a sequence of via points connecting the initial and the final states by applying Dijkstra's minimum cost algorithm. Contrary to most existing methodologies, the computational complexity of this algorithm decreases with an increase in the number and/or the size of the obstacles in the workspace. An interactive computer program has been developed to implement this methodology for a general planar two-link manipulator. This path planning methodology can be applied to any manipulator for which the workspace and the obstacles can be geometrically represented. The algorithm has been applied to some industrial SCARA robots and the results are discussed.