Proposition of On-Line Velocity Scaling Algorithm for Task Space Trajectories

Abstract

A preliminary proposition of a trajectory scaling algorithm is presented. The new algorithm allows for planning a task space trajectory while taking into account the physical capabilities of the manipulator in terms of limits for joint space velocities and accelerations. Contrary to the classic scaling algorithm, which scales uniformly the whole trajectory and has to be executed before the motion starts, the proposed algorithm can be executed online and the scaling is applied only to the part of the planned trajectory. In the proposed algorithm, the planned motion is analyzed in a specified prediction horizon. Theoretical basics of the classic trajectory scaling method are adapted in such a way that the scaling commences only when the analysis detects a violation of joint velocity or acceleration constraints at the prediction horizon. In this paper, the underlying ideas are introduced and discussed, then the core formulae of the algorithms are detailed. An example of scaling a trajectory of a 2-DOF manipulator in the vicinity of a boundary singularity is presented.