Planning near time-optimal trajectories in 3D
Annika E. Mayer, Marcus Sonntag, Oliver Sawodny · 2017 IEEE Conference on Control Technology and Applications (CCTA) · 2017
In context of human-machine-interaction trajectories need to be rerouted or re-planned with reduced dynamical constraints in order to avoid collisions. This requires online feasible algorithm for trajectory planning in space. In this paper a new method for planning time-optimal trajectories subject to dynamic and geometrical constraints is proposed. Distinctively to other work, both spatial and temporal variables are subject to optimization. Further the optimization can be performed on the fly and the computional complexity increases only linear with respect to the total number of waypoints. The dynamic constraints are deduced from safety concerns, it is assumed that those limits are stricter than those originating from the limits of the robot's dynamic, which are therefore neglected. The proposed concept relies on two components, on one hand a projection module which reduces the N-point trajectory planning problem in space to a sequence of 3-point problems in a plane and on the other hand a nonlinear two-phase model predictive control problem to solve the simplified sub-problems. A proof-of-concept shows that trajectories can be planned for a wide range of dynamical constraints in reasonable calculation times.