Online and Robust Intermittent Motion Planning in Dynamic and Changing Environments. Xu, Z., Kontoudis, G., P., & Vamvoudakis, K., G. IEEE Transactions on Neural Networks and Learning Systems, PP:1-15, IEEE, 2023.
doi  abstract   bibtex   19 downloads  
In this paper, we present a real-time kinodynamic motion planning methodology for dynamic environments, denoted as RRT-Q X ∞. We leverage RRT X for global path planning and rapid replanning to produce a set of boundary value problems. A Q-learning optimal controller is proposed for waypoint navigation with completely unknown system dynamics, external disturbances, and intermittent communication. The problem is formulated as a finite-horizon, continuous-time zero-sum game, where the control input is the minimizer, and the worst-case disturbance is the maximizer. To reduce the communication overhead , we allow intermittent transmission of control inputs. Moreover , a relaxed persistence of excitation technique is employed to improve the convergence speed of the Q-learning controller. We provide rigorous Lyapunov-based proofs to guarantee the closed-loop stability of the equilibrium point. The efficacy of the proposed RRT-Q X ∞ is illustrated in several scenarios.
@article{
 title = {Online and Robust Intermittent Motion Planning in Dynamic and Changing Environments},
 type = {article},
 year = {2023},
 pages = {1-15},
 volume = {PP},
 publisher = {IEEE},
 id = {81733efc-1065-3d60-85c6-14d5af8daaa5},
 created = {2023-10-06T16:38:07.419Z},
 file_attached = {true},
 profile_id = {397e406e-8482-37ee-9e30-40cb0af7c9d7},
 last_modified = {2023-10-06T16:39:24.425Z},
 read = {false},
 starred = {false},
 authored = {true},
 confirmed = {true},
 hidden = {false},
 private_publication = {false},
 abstract = {In this paper, we present a real-time kinodynamic motion planning methodology for dynamic environments, denoted as RRT-Q X ∞. We leverage RRT X for global path planning and rapid replanning to produce a set of boundary value problems. A Q-learning optimal controller is proposed for waypoint navigation with completely unknown system dynamics, external disturbances, and intermittent communication. The problem is formulated as a finite-horizon, continuous-time zero-sum game, where the control input is the minimizer, and the worst-case disturbance is the maximizer. To reduce the communication overhead , we allow intermittent transmission of control inputs. Moreover , a relaxed persistence of excitation technique is employed to improve the convergence speed of the Q-learning controller. We provide rigorous Lyapunov-based proofs to guarantee the closed-loop stability of the equilibrium point. The efficacy of the proposed RRT-Q X ∞ is illustrated in several scenarios.},
 bibtype = {article},
 author = {Xu, Zirui and Kontoudis, George P and Vamvoudakis, Kyriakos G},
 doi = {10.1109/TNNLS.2023.3303811},
 journal = {IEEE Transactions on Neural Networks and Learning Systems}
}

Downloads: 19