Agentic Self-Evolutionary Replanning for Embodied Navigation

2026-03-03

AI Summary

This work proposes a Self-Evolving Replanning (SERP) framework to address the limitations of existing embodied navigation approaches that rely on fixed action models and struggle to adapt dynamically to failures in complex environments. SERP introduces a paradigm shift from static to evolving action models by continuously refining them at runtime through recent experience. The framework integrates Graph-of-Thought Chain (GCOT) reasoning to enhance replanning efficiency and combines in-context learning with automatic differentiation (ILAD) to enable adaptive function adjustment and parameter resetting. Furthermore, it leverages large language models (LLMs) for efficient inference over distilled graphs. Experiments demonstrate that SERP significantly improves task success rates across diverse simulated and real-world environments while reducing LLM token consumption, thereby validating its robustness and computational efficiency.

Technology Category

• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Intelligent Robots: Embodied AI

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Failure is inevitable for embodied navigation in complex environments. To enhance the resilience, replanning (RP) is a viable option, where the robot is allowed to fail, but is capable of adjusting plan until success. However, existing RP approaches freeze the ego action model and miss the opportunities to explore better plans by upgrading the robot itself. To address this limitation, we propose Self-Evolutionary RePlanning, or SERP for short, which leads to a paradigm shift from frozen models towards evolving models by run-time learning from recent experiences. In contrast to existing model evolution approaches that often get stuck at predefined static parameters, we introduce agentic self-evolving action model that uses in-context learning with auto-differentiation (ILAD) for adaptive function adjustment and global parameter reset. To achieve token-efficient replanning for SERP, we also propose graph chain-of-thought (GCOT) replanning with large language model (LLM) inference over distilled graphs. Extensive simulation and real-world experiments demonstrate that SERP achieves higher success rate with lower token expenditure over various benchmarks, validating its superior robustness and efficiency across diverse environments.

Problem

• embodied navigation
• replanning
• action model
• self-evolution
• failure recovery

Innovation

• Self-Evolutionary RePlanning
• Agentic Action Model
• In-Context Learning with Auto-Differentiation
• Graph Chain-of-Thought
• Embodied Navigation