Planning Optimal Force-Closure Grasps for Curved Objects

This paper investigates the use of various optimization algorithms for solving the problem of planning optimal three-finger force-closure grasps on curved objects. The optimization aims at minimizing the Q distance function which had been shown in the literature to directly reflect quality of grasps and possess function´s smoothness suitable for optimization processes. Four optimization methods including genetic algorithm, evolution strategy, conjugate gradient method and downhill simplex method are investigated. Experimental results are presented along with efficiency comparison of the selected methods.