Wireless-Guided Indoor Navigation
I co-developed ray-tracing wireless digital-twin and PIRL methods that use simulated propagation as a transferable prior for zero-shot indoor navigation. The robot’s SLAM system supplies the map and robot pose; RF observations and physics-informed priors guide the search policy. This earlier work established the decision-making side of my current research on probabilistic RF localization and robotic sensing.
This figure links posterior RF belief maps with indoor navigation, showing how uncertainty-aware localization can inform a search policy.
The rollout animation shows a navigation policy using wireless belief maps as spatial guidance during indoor search.
Related papers
- Zero-Shot Wireless Indoor Navigation through Physics-Informed Reinforcement Learning (ICRA 2024)
- Digital Twin-Enhanced Wireless Indoor Navigation: Achieving Efficient Environment Sensing with Zero-Shot Reinforcement Learning (IEEE OJ-COMS 2025)
- Reinforcement Learning with Physics-Informed Symbolic Program Priors for Zero-Shot Wireless Indoor Navigation (RLC 2025 Spotlight)