Navigating Discrete Difference Equation Governed WMR by Virtual Linear Leader Guided HMPC

Chao Huang, Xin Chen, Enyi Tang, Mengda He, Lei Bu, Shengchao Qin, Yifeng ZengICRA 2020 · OpenReview

Abstract

In this paper, we revisit model predictive control (MPC) for the classical wheeled mobile robot (WMR) navigation problem. We prove that the reachable set based hierarchical MPC (HMPC), a state-of-the-art MPC, cannot handle WMR navigation in theory due to the non-existence of non-trivial linear system with an under-approximate reachable set of WMR. Nevertheless, we propose a virtual linear leader guided MPC (VLL-MPC) to enable HMPC structure. Different from current HMPCs, we use a virtual linear system with an under-approximate path set rather than the traditional trace set to guide the WMR. We provide a valid construction of the virtual linear leader. We prove the stability of VLL-MPC, and discuss its complexity. In the experiment, we demonstrate the advantage of VLL-MPC empirically by comparing it with NMPC, LMPC and anytime RRT* in several scenarios.
BibTeX
@inproceedings{huang2020navigating,
  author = {Chao Huang and Xin Chen and Enyi Tang and Mengda He and Lei Bu and Shengchao Qin and Yifeng Zeng},
  title = {{Navigating Discrete Difference Equation Governed WMR by Virtual Linear Leader Guided HMPC}},
  booktitle = {IEEE International Conference on Robotics and Automation},
  year = {2020},
  url = {https://openreview.net/forum?id=jjh3bRdrYc1},
}

Loading related content...

Figure