Uncertainty aware maps and risk calls improve indoor robot navigation
DUGM-R: Uncertainty-Aware Dynamic Grid Mapping and Risk-Triggered Recovery for Learned Local Navigation
Robotics
Summary
Navigating robots in crowded indoor spaces can struggle to predict where people and obstacles will move, sometimes causing collisions. The authors created a new way for robots to map their surroundings that includes uncertainty about moving obstacles. Additionally, they trained a backup plan that kicks in when the robot is likely to bump into something, improving safety. They tested this approach in both simulation and on a small robot, showing better navigation and fewer collisions without extra training.
What this means in practice
- •For robotics engineers: Build indoor robots that better predict moving obstacles and avoid collisions by combining uncertainty-aware maps with risk-triggered recovery policies.
- •For logistics automation teams: Deploy navigation systems on delivery or clinical logistics robots that handle crowd dynamics without retraining for each new environment.
Authors
Haoyun Feng, Adrian Rubio-Solis, Zhaodong Guo, George Mylonas
Abstract
Learned local navigation in crowded indoor environments is sensitive to how dynamic obstacle motion is represented, while collision-prone behaviour may persist after nominal policy training. We present a risk-aware reinforcement-learning framework that addresses these two issues through an uncertainty-aware Dynamic Uncertainty Grid Map (DUGM) and a modular post-training recovery mechanism. DUGM combines local occupancy, estimated obstacle motion, and motion-estimation uncertainty in a robot-centric representation. After the nominal policy is frozen, a finite-horizon Risk Value Function (RVF) is trained from nominal rollouts and used to trigger a dedicated recovery policy when continued nominal execution is predicted to be collision-prone. Experiments in a held-out NVIDIA Isaac Sim clinical-logistics benchmark show that uncertainty-aware dynamic representation improves nominal navigation over static and deterministic alternatives, while the recovery mechanism further mitigates residual collision-prone behaviour. The complete framework is also deployed directly on a TurtleBot3 without policy fine-tuning, retraining, or site-specific adaptation, retaining the performance trend observed in simulation. These results indicate that uncertainty-aware dynamic representation and post-training recovery provide complementary mechanisms for improving learned local navigation.