Abstract:

In this paper, we consider an optimization problem of the grasping manipulation
of multi-fingered hand-arm robots. We first formulate an optimization model
for the problem, based on the dynamic equations of the object and the friction constraints.
Then, we reformulate the model as a convex quadratic programming over
circular cones. Moreover, we propose a primal-dual interior-point algorithm based on
the kernel function to solve this convex quadratic programming over circular cones.
We derive both the convergence of the algorithm and the iteration bounds for largeand
small-update methods, respectively. Finally, we carry out the numerical tests of
180◦ and 90◦ manipulations of the hand-arm robot to demonstrate the effectiveness
of the proposed algorithm.