English

Learning Diffusion Policy from Primitive Skills for Robot Manipulation

Robotics 2026-01-06 v1

Abstract

Diffusion policies (DP) have recently shown great promise for generating actions in robotic manipulation. However, existing approaches often rely on global instructions to produce short-term control signals, which can result in misalignment in action generation. We conjecture that the primitive skills, referred to as fine-grained, short-horizon manipulations, such as ``move up'' and ``open the gripper'', provide a more intuitive and effective interface for robot learning. To bridge this gap, we propose SDP, a skill-conditioned DP that integrates interpretable skill learning with conditional action planning. SDP abstracts eight reusable primitive skills across tasks and employs a vision-language model to extract discrete representations from visual observations and language instructions. Based on them, a lightweight router network is designed to assign a desired primitive skill for each state, which helps construct a single-skill policy to generate skill-aligned actions. By decomposing complex tasks into a sequence of primitive skills and selecting a single-skill policy, SDP ensures skill-consistent behavior across diverse tasks. Extensive experiments on two challenging simulation benchmarks and real-world robot deployments demonstrate that SDP consistently outperforms SOTA methods, providing a new paradigm for skill-based robot learning with diffusion policies.

Keywords

Cite

@article{arxiv.2601.01948,
  title  = {Learning Diffusion Policy from Primitive Skills for Robot Manipulation},
  author = {Zhihao Gu and Ming Yang and Difan Zou and Dong Xu},
  journal= {arXiv preprint arXiv:2601.01948},
  year   = {2026}
}

Comments

Accepted to AAAI2026

R2 v1 2026-07-01T08:50:36.996Z