Research paper
AC-SMPC · Learning with navigation state constraints
The method combines actor–critic learning with model predictive control containing explicit navigation constraints. Its collision-avoidance formulation uses local approximations and penalized slack, which make constraint feasibility and residual risk important review questions.
Reviewed 2026-11-05 · Source and bibliography checked
- Original title
- AC-SMPC: Differentiable Actor--Critic MPC with State Constraints for Safe and Fast Robot Navigation
- Publisher
- arXiv
- Authors
- Wu, Chaoran, Fan, Jiayu, Tan, Qichen, Liu, Yuyi, Wang, Jin
- First archive posting
- 2026-10-30
- Source language
- en
- Source country
- US
- Edition
- arXiv:2610.41477v1
- Access
- Open archive metadata and manuscript
- Publication status
- Preprint
Related resources and engineering notes
Source overview
The method combines actor–critic learning with model predictive control containing explicit navigation constraints. Its collision-avoidance formulation uses local approximations and penalized slack, which make constraint feasibility and residual risk important review questions.
Use and review
Source check covers the archive abstract and metadata. Reported results depend on the stated models and tests; application-specific safety validation remains required. The date is the first archive posting.
Sources & editions
AC-SMPC: Differentiable Actor--Critic MPC with State Constraints for Safe and Fast Robot Navigation ↗
arXiv · arXiv:2610.41477v1
arXiv abstract, citation metadata and v1 submission history. First archive posting: 2026-10-30.