OASIS: Optimization-Aware Subgoal Inference with Safety Constraints for Autonomous Mobile Robot Navigation
OASIS: Optimization-Aware Subgoal Inference with Safety Constraints for Autonomous Mobile Robot Navigation
This research introduces OASIS, a hierarchical navigation framework that combines reinforcement learning subgoal generation with model predictive control. The system uses a Soft Actor-Critic policy to generate local subgoals while a nonlinear model predictive controller tracks them under motion limits and collision-avoidance constraints. Optimization outcomes from each subgoal interval are fed back to the policy without additional optimization calls. Evaluated across five navigation scenarios involving static and moving obstacles, OASIS achieves the highest aggregate navigation success compared to hierarchical and direct-goal control variants. Consistent advantages are observed across three training seeds when comparing matched configurations with and without optimization-status input.
The paper presents OASIS, a hierarchical navigation framework that integrates optimization outcomes from model predictive control into reinforcement learning-based subgoal generation to improve success rates in dynamic environments with moving obstacles and sensing uncertainty.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
arXiv:2611.11996v2 Announce Type: replace Abstract: Navigation in dynamic environments requires mobile robots to make progress toward their goals while respecting motion limits and avoiding surrounding obstacles. Learned subgoal policies can guide model predictive control beyond its local prediction horizon, but a geometrically suitable subgoal does not necessarily lead to a successfully solved control problem. This
Open source S2
aints. The outcomes of controller optimizations over each subgoal interval are summarized and provided to the policy at its next decision, without requiring additional optimization calls. The framework is evaluated in five navigation scenarios with static and moving obstacles, including dense interactions and sensing uncertainty. Comparisons with hierarchical and direct-goal control variants show that OASIS achieves
Open source S4
the highest aggregate navigation success. Matched comparisons across three training seeds further show a consistent success advantage over the otherwise identical hierarchy without optimization-status input. The selected-policy comparison also shows fewer primary optimization failures with similar mean controller time.
Open source S5
Source：arXiv Robotics — research abstracts · arxiv.org