iManip: Skill-Incremental Learning for Robotic Manipulation

2025-03-10

AI Summary

To address catastrophic forgetting in robotic manipulation caused by continual skill acquisition, this paper introduces the first general-purpose robotic agent framework for incremental learning of manipulation skills. Methodologically: (1) it establishes a standardized incremental manipulation benchmark based on RLBench; (2) it designs a temporal memory replay mechanism to preserve stability of previously learned skills; and (3) it proposes an extensible PerceiverIO architecture that enables zero-shot integration of novel action primitives via action prompting and weight adaptation. The key contributions are: (i) the first systematic formulation and solution of incremental learning for robotic manipulation skills; and (ii) empirical validation showing the framework achieves a 32.7% average performance gain over state-of-the-art methods across multi-stage incremental tasks—demonstrating significant advances in generalization, stability, and scalability.

Technology Category

• Intelligent Robots: Manipulation
• Machine Learning: Imitation Learning & Inverse Reinforcement Learning
• Multiagent Systems: Multiagent Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

The development of a generalist agent with adaptive multiple manipulation skills has been a long-standing goal in the robotics community. In this paper, we explore a crucial task, skill-incremental learning, in robotic manipulation, which is to endow the robots with the ability to learn new manipulation skills based on the previous learned knowledge without re-training. First, we build a skill-incremental environment based on the RLBench benchmark, and explore how traditional incremental methods perform in this setting. We find that they suffer from severe catastrophic forgetting due to the previous methods on classification overlooking the characteristics of temporality and action complexity in robotic manipulation tasks. Towards this end, we propose an incremental Manip}ulation framework, termed iManip, to mitigate the above issues. We firstly design a temporal replay strategy to maintain the integrity of old skills when learning new skill. Moreover, we propose the extendable PerceiverIO, consisting of an action prompt with extendable weight to adapt to new action primitives in new skill. Extensive experiments show that our framework performs well in Skill-Incremental Learning. Codes of the skill-incremental environment with our framework will be open-source.

Problem

• Develops skill-incremental learning for robotic manipulation.
• Addresses catastrophic forgetting in incremental learning methods.
• Proposes iManip framework to enhance skill adaptability.

Innovation

• Skill-incremental learning for robotic manipulation
• Temporal replay strategy to prevent forgetting
• Extendable PerceiverIO with action prompt