Real-time trajectory planning for systems subject to high order kinematic and dynamic constraints

This work proposes an extended study on real-time trajectory planners for mechatronic systems. Designed planners are based on the so-called path-velocity decomposition paradigm, i.e. trajectories are obtained by first designing the desired geometric path and, then, by associating a time-law to the movement along it. In particular, the first proposed real-time trajectory planner (named Trajectory Scaling System) acts in the configuration space and is able to scale down the longitudinal velocity in order to fulfill a given set of constraints associated to the mechatronic system. More precisely, it manages, in real-time, kinematic constraints on joint velocities, accelerations, and jerks as well as dynamic constraints on generalized joint forces and their derivatives. The same planning scheme has been successively applied to trajectories defined in the operational space. In this case, the planner handles in real-time kinematic bounds on joint velocities, accelerations and jerks. The scaling system has been validated through experimental tests performed on an actual 6 degrees of freedom manipulator by executing a set of different Cartesian paths passing close to singular points. The last considered planning problem concerns the real-time end-effector orientation modification made in order to maintain an exact tracking of the planned time-law. Even in this case, the planning scheme has been validated through experimental tests executed on an anthropomorphic manipulator by performing Cartesian paths passing close to, or even crossing, kinematic singularities.