TR2014-046

Energy-Efficient Collision-Free Trajectory Planning Using Alternating Quadratic Programming


    •  Zhao, Y., Wang, Y., Bortoff, S.A., Nikovski, D., "Energy-efficient Collision-free Trajectory Planning Using Alternating Quadratic Programming", American Control Conference (ACC), DOI: 10.1109/​ACC.2014.6859076, June 2014, pp. 1249-1254.
      BibTeX TR2014-046 PDF
      • @inproceedings{Zhao2014jun,
      • author = {Zhao, Y. and Wang, Y. and Bortoff, S.A. and Nikovski, D.},
      • title = {Energy-efficient Collision-free Trajectory Planning Using Alternating Quadratic Programming},
      • booktitle = {American Control Conference (ACC)},
      • year = 2014,
      • pages = {1249--1254},
      • month = jun,
      • publisher = {IEEE},
      • doi = {10.1109/ACC.2014.6859076},
      • issn = {0743-1619},
      • isbn = {978-1-4799-3272-6},
      • url = {https://www.merl.com/publications/TR2014-046}
      • }
  • MERL Contacts:
  • Research Areas:

    Control, Optimization, Dynamical Systems

Abstract:

This paper considers the planning of collision-free and energy-optimal trajectories for linear systems with decoupled dynamics for different degrees of freedom. A direct transcription of such a problem generally results in a non-convex problem due to the collision avoidance constraint. In this paper we propose a novel Alternating Quadratic Programming (AQP) algorithm to deal with the non-convex collision avoidance constraint, and generate a suboptimal solution for the original problem by alternatively solving a number of subproblems. It is proved that the AQP algorithm is guaranteed to converge, and the solution is locally optimal when the boundary of the collision-free region satisfies certain properties. The speed and energy-saving performance of the proposed method are demonstrated by numerical examples.