Application of a modified Genetic Algorithm for enhancing grasp quality on 3D objects

Robots are increasingly being inducted to perform a variety of complex handling operations in various industries. In contrast to use of parallel jaw grippers, automated grasping by robot fingers, offer more challenges in synthesis and task planning. In this paper, a meta-heuristic optimization method based on a modified Genetic Algorithm (GA) has been formulated for automated synthesis of high quality grasps. This unique modified GA scheme is applied on initially feasible grasps. The widely used largest ball criterion is employed to calculate the quality of the resulting grasps. The performance of the method is numerically presented by the use of tessellated 3D objects for the implementation of the algorithm. The optimization for these test cases is conducted for both frictional and non-frictional cases.