ECo-MoE: Embodiment-Conditioned Mixture of Experts Increases the Evolvability of Robots

ICML 2026

Yibin Wang, Muhan Li, Zihan Guo, Sam Kriegman

Northwestern University

Paper Code Citation

TL;DR: We introduce a novel abstraction and operationalization of evolution that co-optimizes latent-gated mixtures of experts and designs.

Evo on Locomotion Tasks

Flat Ground

Initial-generation robot for the Flat Ground task
Initial Generation
Final-generation robot for the Flat Ground task
Final Generation

Upright Locomotion

Initial-generation robot for the Upright Locomotion task
Initial Generation
Final-generation robot for the Upright Locomotion task
Final Generation

Potholes

Initial-generation robot for the Potholes task
Initial Generation
Final-generation robot for the Potholes task
Final Generation

Evo by Demo

Predefined ant 8-DOF demo robot
Predefined Demo
Evolved robot from evo by demo
Evolved Robot

Abstract

In this paper, we introduce a model of evolution and learning in robots that co-optimizes a distribution of latent design vectors (genotypes) and a mixture of control experts (neural modules), which are gated by the latent coordinates of each decoded design (phenotype). This provides a scalable alternative to co-design algorithms that either train an individual policy for every robot, which is inefficient, or a monolithic universal controller for all robots, which results in overly conservative structures and behaviors. Our approach lies somewhere between these two extremes, preserving ancestral knowledge in a unified yet modular framework in which different body plans activate and deactivate different combinations of learned sensorimotor circuits for goal-directed behavior. This allows one part of the controller to be overhauled to better suit new species of designs as they emerge without disrupting the hard-earned knowledge contained within other expert modules. It also allows pretrained expert policies to be directly plugged into the mixture, which can steer evolution into otherwise unexplored areas of latent space containing desired morphological traits. We refer to this process as "evo by demo" and explore how it may be used to guide freeform evolution toward canonical structures defined by the pretrained model.

Method

Overview of the ECo-MoE method
Embodiment-conditioned mixture of experts. Designs were sampled from an evolving distribution within a latent space of possible genotypes (A and B). The distribution was initialized randomly for the main experiment (blue region in B); for "evo by demo", it was regularized by a predesigned demo (orange region in B). The latent genotype of each endoskeletal phenotype (C and D) was fed as input to a gating network that produces a weighted policy output π(at | st, z) by mixing expert actions (E). In evo-by-demo, a policy was pretrained for the predesigned demo and injected as a frozen expert into the mixture. Designs with latent genes similar to the predefined species were routed with greater weight to the pretrained expert (orange bar in D), steering evolution toward the desired phenotypic traits.

Citation

@inproceedings{wang2026ecomoe,
  title     = {ECo-MoE: Embodiment-Conditioned Mixture of Experts Increases the Evolvability of Robots},
  author    = {Yibin Wang and Muhan Li and Zihan Guo and Sam Kriegman},
  booktitle = {Proceedings of the International Conference on Machine Learning (ICML)},
  year      = {2026}
}