RoboRouter: Training-Free Policy Routing for Robotic Manipulation

2026-03-09

AI Summary

While existing robotic manipulation policies demonstrate effectiveness on specific tasks, their generalization capabilities remain limited. To address this challenge, this work proposes the first training-free policy routing framework that dynamically selects the most suitable policy from a heterogeneous policy pool by leveraging semantic task representations and retrieving relevant historical tasks, thereby enabling zero-shot, lightweight policy integration. The approach incorporates a structured feedback mechanism that effectively fuses experiences from multiple policy sources. Evaluated in both simulation and real-world environments, the method achieves average success rate improvements exceeding 3% and 13%, respectively, significantly enhancing system generalization while maintaining computational efficiency and fast execution.

Technology Category

• Intelligent Robots: Manipulation
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling
• Search and Optimization: Learning to Search

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
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Research on robotic manipulation has developed a diverse set of policy paradigms, including vision-language-action (VLA) models, vision-action (VA) policies, and code-based compositional approaches. Concrete policies typically attain high success rates on specific task distributions but lim-ited generalization beyond it. Rather than proposing an other monolithic policy, we propose to leverage the complementary strengths of existing approaches through intelligent policy routing. We introduce RoboRouter, a training-free framework that maintains a pool of heterogeneous policies and learns to select the best-performing policy for each task through accumulated execution experience. Given a new task, RoboRouter constructs a semantic task representation, retrieves historical records of similar tasks, predicts the optimal policy choice without requiring trial-and-error, and incorporates structured feedback to refine subsequent routing decisions. Integrating a new policy into the system requires only lightweight evaluation and incurs no training overhead. Across simulation benchmark and real-world evaluations, RoboRouter consistently outperforms than in-dividual policies, improving average success rate by more than 3% in simulation and over 13% in real-world settings, while preserving execution efficiency. Our results demonstrate that intelligent routing across heterogeneous, off-the-shelf policies provides a practical and scalable pathway toward building more capable robotic systems.

Problem

• policy routing
• robotic manipulation
• heterogeneous policies
• generalization
• training-free

Innovation

• training-free
• policy routing
• robotic manipulation
• heterogeneous policies
• semantic task representation