mlsanityData to 14 Oct 2026

Shared Modular Recurrence in Contextual MDPs for Universal Morphology Control

arXiv 2506.08630 ↗First posted 10 Jun 2025cs.AI cs.ROcited by 3

Not on the map. Why?

Abstract (arXiv)

A universal controller for any robot morphology would greatly improve computational and data efficiency. Steps have been made towards such multi-robot control by utilizing contextual information about the properties of individual robots and exploiting their modular structure in the architecture of deep reinforcement learning agents. When the robots have highly dissimilar morphologies, however, this becomes a challenging problem, especially when the agent must generalize to new, unseen robots. In this paper, we posit that contextual features are often only partially available, but that they can be recovered through modular interactions. This can allow for better multi-robot control and generalization to contexts that are not seen during training. To this extent, we implement a transformer-based architecture with shared modular recurrence and evaluate its (generalization) performance on a large set of MuJoCo robots. The results show a substantial improvement in zero-shot generalization performance on robots with unseen dynamics, kinematics, and topologies, in four different environments.

Key facts. Cited by 3 papers in the map (data through 14 Oct 2026). Builds on 28 of its 37 references that are in the map. First posted on arXiv 10 Jun 2025.

Builds on

28 of 37 references.

References · 28

  1. 2017Attention Is All You Needcited by 60,361
  2. 2017Proximal Policy Optimization Algorithmscited by 15,215
  3. 2023RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Controlcited by 3,080
  4. 2022RT-1: Robotics Transformer for Real-World Control at Scalecited by 2,061
  5. 2024Octo: An Open-Source Generalist Robot Policycited by 1,484
  6. 2014On the Properties of Neural Machine Translation: Encoder-Decoder Approachescited by 1,304
  7. 2016HyperNetworkscited by 1,027
  8. 2015End-to-End Training of Deep Visuomotor Policiescited by 875
  9. 2023Open X-Embodiment: Robotic Learning Datasets and RT-X Modelscited by 862
  10. 2018QT-Opt: Scalable Deep Reinforcement Learning for Vision-Based Robotic Manipulationcited by 743
  11. 2017Sim-to-Real Transfer of Robotic Control with Dynamics Randomizationcited by 672
  12. 2015Deep Recurrent Q-Learning for Partially Observable MDPscited by 465
16 more
  1. 2015Contextual Markov Decision Processescited by 220
  2. 2021A Survey of Zero-shot Generalisation in Deep Reinforcement Learningcited by 170
  3. 2021Embodied Intelligence via Learning and Evolutioncited by 163
  4. 2017Preparing for the Unknown: Learning a Universal Policy with Online System Identificationcited by 141
  5. 2020One Policy to Control Them All: Shared Modular Policies for Agent-Agnostic Controlcited by 121
  6. 2024Scaling Cross-Embodied Learning: One Policy for Manipulation, Navigation, Locomotion and Aviationcited by 114
  7. 2021Why Generalization in RL is Difficult: Epistemic POMDPs and Implicit Partial Observabilitycited by 109
  8. 2022MetaMorph: Learning Universal Controllers with Transformerscited by 89
  9. 2019Learning to Control Self-Assembling Morphologies: A Study of Generalization via Modularitycited by 70
  10. 2020My Body is a Cage: the Role of Morphology in Graph-Based Incompatible Controlcited by 70
  11. 2021The Sensory Neuron as a Transformer: Permutation-Invariant Neural Networks for Reinforcement Learningcited by 50
  12. 2024One Policy to Run Them All: an End-to-end Learning Approach to Multi-Embodiment Locomotioncited by 43
  13. 2024Body Transformer: Leveraging Robot Embodiment for Policy Learningcited by 32
  14. 2024GET-Zero: Graph Embodiment Transformer for Zero-shot Embodiment Generalizationcited by 27
  15. 2023Universal Morphology Control via Contextual Modulationcited by 23
  16. 2024Distilling Morphology-Conditioned Hypernetworks for Efficient Universal Morphology Controlcited by 11

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 3.

3 papers.

  1. 2026Contextual Intelligence The Next Leap for Reinforcement Learningcited by 1
  2. 2026Dynamic Context Scheduling: Learning Beyond the Static Universecited by 0
  3. 2026RecMorph: Topology-Guided Spatial Recurrence for Generalized Morphology Controlcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025GCNT: Graph-Based Transformer Policies for Morphology-Agnostic Reinforcement Learning
  2. 2024DittoGym: Learning to Control Soft Shape-Shifting Robots
  3. 2025McARL:Morphology-Control-Aware Reinforcement Learning for Generalizable Quadrupedal Locomotion
  4. 2024Towards Multi-Morphology Controllers with Diversity and Knowledge Distillation
  5. 2025Accelerated co-design of robots through morphological pretraining
  6. 2026Graph-Operator World Models for Morphology-Parameter Generalization in Continuous Control
  7. 2026Embedding Morphology into Transformers for Cross-Robot Policy Learning
  8. 2025Robot Control Stack: A Lean Ecosystem for Robot Learning at Scale