Genetic algorithm based direction finder on the manifold for singularity free paths

Manipulators that execute tasks by following assigned trajectories, experience limitations in mobility when encountering singularities within their workspace. Generating trajectories for manipulators thus requires the avoidance of singular configurations as way points of the path to prevent the manipulator from assuming poses that limit the degrees of freedom of the manipulator. This paper establishes the need to operate in a singularity free configuration space, utilizes the conditions to define a singularity free manifold for a non-redundant, holonomic manipulator and proposes the Genetic Algorithm Based Direction Finder (GADF) to determine the optimal directions of motion on the implicitly defined manifold to generate a singularity free path from the start to the goal configurations of the defined task. The GADF was successfully simulated using a model of a 3 link planar manipulator followed by an analysis of results and possible improvements.