Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
March 21, 2026International Journal of Robust and Nonlinear Control

Safe Reinforcement Learning for Optimal Tracking of Continuous‐Time Nonlinear Systems

View Full Paper
Ask AI
Bookmark
Share

Authors

SKSoha KansoMJMayank Shekhar JhaDTDidier Theilliol

Discussion

Loading...

Member takes

Overview

This work develops a safe reinforcement learning approach to optimize tracking in nonlinear systems, suggesting improved performance and safety.

Key Points

  • The study aims to develop a safe reinforcement learning method for optimal tracking of continuous-time nonlinear systems while ensuring safety.
  • Synthesize an optimal tracker under safety constraints using an off-policy approach.
  • Formulate an augmented system comprising tracking error and state dynamics.
  • Apply quadratic programming and control barrier functions during exploration and exploitation phases.
  • Utilize neural networks to approximate the optimal control law.
  • Demonstrated the ability to maintain safety while achieving optimal tracking performance.
  • Innovative mathematical proofs ensured safety, stability, and convergence to optimal solutions.
  • Validation through simulation showed effectiveness of the proposed approach.

Cite This Study

Kanso et al. (2026) studied this question.

synapsesocial.com/papers/69be36416e48c4981c675146https://doi.org/10.1002/rnc.70518
View Full Paper
Ask AI
Bookmark
Share