# Skill Composition for Legged Robot Reinforcement Learning

arXiv: https://arxiv.org/abs/2609.14647 · HTML: https://mlsanity.com/paper/2609.14647/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in RL for robot locomotion and manipulation, one of 15 directions in Deep reinforcement learning methods.
- Builds on 19 of its 24 references that are in the map.
- First posted on arXiv 13 Sep 2026.

## Placed in

- Deep reinforcement learning methods / [RL for robot locomotion and manipulation](https://mlsanity.com/deeprl/rl-for-robot-locomotion-and-manipulation/)

## Builds on (most cited first, 19 of 19)

- [OpenVLA: An Open-Source Vision-Language-Action Model](https://mlsanity.com/paper/2406.09246/) (2024)
- [Do As I Can, Not As I Say: Grounding Language in Robotic Affordances](https://mlsanity.com/paper/2204.01691/) (2022)
- [Learning Quadrupedal Locomotion over Challenging Terrain](https://mlsanity.com/paper/2010.11251/) (2020)
- [Learning Invariant Representations for Reinforcement Learning without Reconstruction](https://mlsanity.com/paper/2006.10742/) (2020)
- [Residual Reinforcement Learning for Robot Control](https://mlsanity.com/paper/1812.03201/) (2018)
- [Robust Motion In-betweening](https://mlsanity.com/paper/2102.04942/) (2021)
- [Contextual Markov Decision Processes](https://mlsanity.com/paper/1502.02259/) (2015)
- [Humanoid Parkour Learning](https://mlsanity.com/paper/2406.10759/) (2024)
- [Sequential Dexterity: Chaining Dexterous Policies for Long-Horizon Manipulation](https://mlsanity.com/paper/2309.00987/) (2023)
- [Adversarial Skill Chaining for Long-Horizon Robot Manipulation via Terminal State Regularization](https://mlsanity.com/paper/2111.07999/) (2021)
- [Perceptive Humanoid Parkour: Chaining Dynamic Human Skills via Motion Matching](https://mlsanity.com/paper/2602.15827/) (2026)
- [Cascaded Compositional Residual Learning for Complex Interactive Behaviors](https://mlsanity.com/paper/2212.08954/) (2022)
- [SkillBlender: Towards Versatile Humanoid Whole-Body Loco-Manipulation via Skill Blending](https://mlsanity.com/paper/2506.09366/) (2025)
- [Training Transition Policies via Distribution Matching for Complex Tasks](https://mlsanity.com/paper/2110.04357/) (2021)
- [N2M: Bridging Navigation and Manipulation by Learning Pose Preference from Rollout](https://mlsanity.com/paper/2509.18671/) (2025)
- [Learning Agile Striker Skills for Humanoid Soccer Robots from Noisy Sensory Input](https://mlsanity.com/paper/2512.06571/) (2025)
- [Discovery of skill switching criteria for learning agile quadruped locomotion](https://mlsanity.com/paper/2502.06676/) (2025)
- [RPG: Robust Policy Gating for Smooth Multi-Skill Transitions in Humanoid Fighting](https://mlsanity.com/paper/2604.21355/) (2026)
- [Learning Setup Policies: Reliable Transition Between Locomotion Behaviours](https://mlsanity.com/paper/2101.09391/) (2021)

## Abstract (arXiv)

Robots, and humanoid robots in particular, are increasingly competent at individual behaviors, each obtained by training a specialized controller. A specialized skill is quick to train, converges reliably because the problem it faces is narrow, and can be validated on its own, none of which is true of a single end-to-end policy asked to cover everything. What remains fragile is the transition between them. We argue that the composition of independent sub-policies deserves to be treated as a research problem in its own right, rather than as an implementation detail left to whatever mechanism happens to be at hand. Reliable composition is what turns a collection of separate skills into a repertoire that can be used, extended and shared. More fundamentally, if control can be passed between specialized policies safely, and at any moment, the choice of what the robot should do next can be delegated to a component of an entirely different nature, such as a planner, an automaton or a symbolic controller, whose behavior can be inspected in advance. The policies would then only ever have to act, and what the robot can be trusted to do would become verifiable.