Lifelong Embodied Navigation Learning

2026-03-06

AI Summary

This work addresses the challenge of catastrophic forgetting in embodied navigation agents when continuously learning across multiple environments and instruction-following tasks. To this end, it formally defines the lifelong learning problem in embodied navigation for the first time and introduces Uni-Walker, a novel framework that decouples navigation knowledge into task-shared and task-specific components. Uni-Walker integrates Decoder Extension LoRA (DE-LoRA), knowledge inheritance, expert co-activation, subspace orthogonality constraints, and a navigation-specific chain-of-thought mechanism to effectively balance the acquisition of new skills with the retention of previously learned knowledge. Experimental results demonstrate that Uni-Walker significantly outperforms existing approaches, offering a promising pathway toward building general-purpose embodied navigation agents with robust lifelong learning capabilities.

Technology Category

• Machine Learning: Life-Long and Continual Learning
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling
• Intelligent Robots: Embodied AI

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Embodied navigation agents powered by large language models have shown strong performance on individual tasks but struggle to continually acquire new navigation skills, which suffer from catastrophic forgetting. We formalize this challenge as lifelong embodied navigation learning (LENL), where an agent is required to adapt to a sequence of navigation tasks spanning multiple scenes and diverse user instruction styles, while retaining previously learned knowledge. To tackle this problem, we propose Uni-Walker, a lifelong embodied navigation framework that decouples navigation knowledge into task-shared and task-specific components with Decoder Extension LoRA (DE-LoRA). To learn the shared knowledge, we design a knowledge inheritance strategy and an experts co-activation strategy to facilitate shared knowledge transfer and refinement across multiple navigation tasks. To learn the specific knowledge, we propose an expert subspace orthogonality constraint together and a navigation-specific chain-of-thought reasoning mechanism to capture specific knowledge and enhance instruction-style understanding. Extensive experiments demonstrate the superiority of Uni-Walker for building universal navigation agents with lifelong learning.

Problem

• lifelong learning
• embodied navigation
• catastrophic forgetting
• navigation tasks
• knowledge retention

Innovation

• lifelong learning
• embodied navigation
• knowledge decoupling
• LoRA
• catastrophic forgetting