Using biased randomization for trajectory optimization in robotic manipulators

Abstract

We study the problem of optimization of trajectories for a robotic manipulator, with two degrees of freedom, which is constrained to pass through a set of waypoints in the workspace. The aim is to determine the optimal sequence of points and continuous optimal system trajectory. The actual formulation involves an optimal control problem of a dynamic system within integer variables that model the waypoints constrains. The nature of this problem, highly nonlinear and combinatorial, makes it particularly difficult to solve. The proposed method combines a meta-heuristic algorithm to determine the promising sequence of discrete points with a collocation technique to optimize the continuous path of the system. This method does not guarantee the global optimum, but can solve instances of dozens of points in reasonable computation time.