Planning Optimal Force-Closure Grasps for Curved Objects by Genetic Algorithm

This paper describes the use of genetic algorithm to solve the problem of planning optimal force-closure grasps on 2D and 3D objects. The proposed approach searches for optimal grasps from parametric curves of a 2D object or parametric surfaces of a 3D object. The search method is based on the concept of Q distance which can guide the search direction to better solutions. On this metric, genetic algorithm can exploit the characteristic of Q distance to find optimal grasps by directly using the computation of Q distance as the fitness function. This approach is applied to plan optimal force-closure grasps on 2D and 3D curved objects with three and four contact points respectively. Preliminary results are also presented