LAUSR

The problem of automated grasp generation is exacerbated by the infinite types of objects to be handled by robots. In this work, the issue is cast as an optimization problem and a modified genetic algorithm-based approach has been formulated for the synthesis of high-quality grasps. The convex hull of the grasp contact wrenches is built, and the largest ball is inscribed within it. The radius of this resulting ball, centered at the origin, is used to represent the grasp quality. An initial feasible grasp is increased in quality by generating wrench population considering the complete body for an exhaustive search. Tessellated objects are utilized for the planner to ensure the applicability of the approach on complex shapes. The performance efficacy of the proposed method is numerically showcased through various frictional and non-frictional prehensile contact examples and is featured along with the results of an existing heuristic method on similar models with moderate and dense tessellation.