Duke's General Robotics Lab open-sources RTNav for real-time zero-shot object navigation
rsasaki0109 · x · 2026-10-13
Duke University's General Robotics Lab (Easop Lee, Lingyu Zhang, Boyuan Chen) released RTNav, tackling zero-shot object navigation in real time: unlike standard setups, the simulator keeps stepping at a fixed frequency while the agent computes actions, forcing decisions in a moving world.
- The repo ships a modular, asynchronous navigation agent plus a unified evaluation framework supporting 6 baselines, built on a Habitat simulator running at 30 Hz in Docker, with each agent in its own container.
- Accompanying figures contrast synchronous (world paused during inference) vs. asynchronous execution and plot baselines on a decision-latency vs. performance-degradation curve.
- A separate RTNav Real-World repo targets physical deployment.
Related event: Duke Open-Sources RTNav for Real-Time Zero-Shot Object Navigation(2 posts)→
More from Embodied
- 7 AI models flew a drone via phone MCP: Claude refused, Codex one-shotted it — Basic-Let6828 · 2026-10-13
- SEA-LLM from a Google internship brings egocentric spatial audio understanding to smart glasses — AcouIntel · 2026-10-13
- Proposing a 'Millennium Problem of Robotics': ten long-horizon research challenges listed — DJiafei · 2026-10-13
- DynaRobotics trains whole-body RL controller for Taku as labs converge on RL+distill — chris_j_paxton · 2026-10-13
- Open-source wearable motorless leader arm captures teleoperation data for robot learning — chris_j_paxton · 2026-10-13
- Apple launches HomePod mini 2 with S10 chip and Sound Glow at $149 — Scobleizer · 2026-10-13