RecMorph improves limb coordination in robots with many parts
RecMorph: Topology-Guided Spatial Recurrence for Generalized Morphology Control
RoboticsMachine Learning
Summary
Controlling robots with many limbs is tricky because each limb can be very different and all need to work together smoothly. The authors created RecMorph, a system that helps information move across all parts of a robot more efficiently by following the robot’s body structure. This system can handle bigger robots with up to 30 limbs and works well both in simulations and on real robots. By using this structure, RecMorph improves how a single controller manages the whole robot’s movement, even in challenging conditions.
What this means in practice
- •For robotics engineers: Use RecMorph to improve control policies for robots with many joints and limbs, enabling better whole-body coordination and robustness in various environments.
- •For robotics platform developers: Deploy a single learned policy across different multi-limbed robot platforms, reducing the effort required to adapt motion controllers to new robot shapes.
Authors
Quanrui Rao, Yong Liu, Xueming Xiao, Yingbo Luo, Kun Wu, Zhenyu Xu, Meibao Yao
Abstract
Generalized morphology control requires a single policy to transform information across limbs with different physical roles, coordinate whole-body motion, and remain efficient as body size grows. Existing communication mechanisms address these requirements only partially. We introduce RecMorph, a topology-guided spatial recurrent architecture that uses recurrent sequence computation to jointly perform cross-limb communication and representation transformation. A depth-first traversal converts the kinematic tree into a morphology-derived sequence, along which shared bidirectional transitions progressively transform limb information before action decoding. Residual preservation, RMS normalization, and input-dependent channel modulation stabilize this repeated spatial transformation, yielding linear token complexity at fixed model width and depth. Across five UNIMAL tasks, RecMorph achieves the strongest mean final training performance among the evaluated generalized morphology controllers and the highest measured inference throughput on FT, while generalizing to unseen variations and bodies with up to 30 limbs. We further migrate representative generalized controllers from UNIMAL benchmarks to a four-platform quadruped setting. RecMorph achieves the best macro-averaged performance under nominal and high friction, reduces nominal velocity RMSE by 43.5% relative to specialist MLPs, and one shared policy completes 40 physical Go1/Go2 trials without falls. These results show that topology-guided recurrent transformation provides an effective and efficient communication mechanism for Generalized Morphology Control and remains effective when transferred from procedural bodies to physical robot platforms. Code and experimental resources are publicly available at https://github.com/quanruirao/RecMorph.