An experimentally validated technique for the real-time management of wrist singularities in nonredundant anthropomorphic manipulators

The automatic management of kinematic singularities, which are typical for trajectories planned in the operational space, is arousing a renewed interest among the scientific community because the most recent strategies make it possible their real-time management. The approach described in this paper allows executing trajectories in the operational space which pass through wrist singularities. It introduces several novelties w.r.t. known alternative strategies. First of all, it is conceived for trajectories which are planned on-the-fly. Secondly, singularities are avoided by changing slightly the tool-frame orientation while strictly preserving both the assigned Cartesian path and time-law. Finally, the approach is effective also for manipulators moving at standard operative speeds and it explicitly handles given limits on joint velocities and accelerations. In this paper an approach proposed in early works is revised in order to make it ready for an industrial implementation. In particular a procedural method is proposed for the tuning of the algorithm, so as to make it more deterministic and to increase the success rates. Furthermore, the singularity avoidance problem is theoretically analyzed in order to devise a necessary condition for the the existence of a solution. Results are experimentally validated through an anthropomorphic industrial manipulator.