Navigation agents struggle when environments change during tasks
EvoNav-Bench: Benchmarking Lifelong Navigation in Evolving Environments
Robotics
Summary
Navigation programs often learn to move through a place by remembering what they saw before. But if the place changes, like furniture moving or doors opening, their old memories can cause mistakes. The authors made a new test setting where the environment changes between tasks to see how different navigation methods handle this. They found that current fancy methods aren't good at dealing with changes, but simple tricks to update or reset memory can help. This work helps us test robots or apps that need to move through places that aren’t always the same.
What this means in practice
- •For robotics engineers: Test and improve robot navigation systems that must handle environments changing over time.
- •For video game developers: Evaluate NPC navigation that adapts when game worlds are modified between player interactions.
Authors
Xilin Wang, Guoxi Zhang, Hongming Xu, Zhuofan Zhang, Tianxu Wang, Lifeng Fan
Abstract
Lifelong navigation (LN) requires an embodied agent to solve a sequence of navigation subtasks in the same environment. Since solving each subtask from scratch incurs redundant exploration, an LN agent must consolidate experience from earlier stages and reuse it in later stages, often through persistent scene representations such as scene graphs or visual snapshots. However, existing approaches typically assume a stationary environment, whereas in real-world LN settings, human activities can cause the environment to evolve. With the stationary assumption violated, existing methods may fuse outdated prior observations with new observations, yet current benchmarks cannot reveal this failure mode. In this paper, we present EvoNav-Bench, which extends the GOAT-Bench style LN formulation in the context of evolving environments. Built on the ProcTHOR framework, EvoNav-Bench introduces environment modifications between navigation tasks, making prior experience useful but not fully reliable. This design enables controlled evaluation of how environment evolution affects LN agents that reuse prior scene observations. Using EvoNav-Bench, we benchmark three recent methods that build and reuse scene representations for navigation. We also compare three simple heuristic strategies for handling environment evolution: Frontier-Update, Fail-then-Update, and Stage-Reset. Our results show that existing methods are brittle under environment evolution, while the heuristic strategies enable a controlled analysis of how agents can adapt to scene changes and mitigate their impact.