Avalan enables indoor robot navigation without prior site mapping
AVALON: Zero-Shot Indoor Navigation with Human-Seeded Topological Memory
RoboticsArtificial Intelligence
Summary
Indoor robots often struggle to navigate buildings they haven't seen before. The authors present AVALON, a method that uses human-provided hints and a basic floor plan to help robots find their way without creating detailed maps first. The robot checks its location using visual clues and adjusts its path as needed, while humans and robots update a shared memory to handle changes or task needs. This approach was tested on a real robot and showed faster, more reliable navigation compared to other methods that rely on building detailed maps.
What this means in practice
- •For service robot developers: Enable robots to navigate new indoor environments quickly using only floor plans and human hints without building detailed maps first.
- •For facility management teams: Use robots to perform indoor tasks efficiently by updating shared memory of environmental changes through coordinated human and robot input.
Authors
Jing Xia, Changyan Li, Yuduo Wang, Guiyin Tian, Bo Wang, Yifan Liao, Mingjun Li, Shiyan Hu
Abstract
Indoor service robots often enter unmapped buildings whose layouts and task constraints are already familiar to users. We address this human-familiar, robot-unseen setting with AVALON, a topological navigation framework organized around human-seeded layered spatial memory, node-verified topological execution, and task-triggered, evidence-gated human--robot memory co-updating. The robot has no prior on-site traversal or site-specific training. Deployment requires only a floor plan, place clues, and an initial topology-odometry alignment. Visual depth point-goal navigation leverages verified node arrivals to perform translational re-anchoring, without requiring a robot-constructed global SLAM map. Human--robot memory co-updating fuses complementary sources of information. Human operators specify task constraints that are unobservable via vision and report offline environmental modifications, whereas the robot provides physical-state evidence through active in-situ validation. Task relevance governs whether human-reported environmental changes undergo robot validation during the active mission. Field-collected perceptual evidence gates memory updates, maintaining an explicit separation between human-specified permissions and robot-observed traversability. Our experiments on a Unitree Go2 show that AVALON with X-NavDP completes a 12.81 m Intersection-to-Dining mission in 46.4 s (5/5) with no mapping, 40.9% less robot time than DimOS-Go2 (78.5 s, 4/5), while policy-only X-NavDP times out in all of the 5 runs. These demonstrate the superiority of the proposed topological navigation framework.