Controller Distillation Reduces Fragile Brain-Body Co-Adaptation and Enables Migrations in MAP-Elites

2025-04-09

AI Summary

This work addresses the failure of controller transfer across morphologically distinct bodies in brain-body coevolution, caused by excessive controller specialization to specific morphologies. We identify fragile brain-body co-adaptation as the root cause of impaired solution transfer in MAP-Elites. To overcome this, we propose Pollination Distillation: a knowledge distillation mechanism that decouples morphological mutation from controller optimization, thereby enhancing controller generalizability; combined with morphology-feature-driven niche partitioning and evolutionary transfer assessment, it improves effective solution migration across morphological subspaces in quality diversity (QD) search. Experiments demonstrate substantial gains—average QD score improvements of 27% across multi-robot morphology tasks—along with significantly higher transfer success rates and elite coverage. This constitutes the first systematic validation of feasible and effective cross-morphology controller reuse.

Technology Category

• Search and Optimization: Evolutionary Computation
• Intelligent Robots: Manipulation
• Machine Learning: Evolutionary Learning

Application Category

• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Brain-body co-optimization suffers from fragile co-adaptation where brains become over-specialized for particular bodies, hindering their ability to transfer well to others. Evolutionary algorithms tend to discard such low-performing solutions, eliminating promising morphologies. Previous work considered applying MAP-Elites, where niche descriptors are based on morphological features, to promote better search over morphology space. In this work, we show that this approach still suffers from fragile co-adaptation: where a core mechanism of MAP-Elites, creating stepping stones through solutions that migrate from one niche to another, is disrupted. We suggest that this disruption occurs because the body mutations that move an offspring to a new morphological niche break the robots' fragile brain-body co-adaptation and thus significantly decrease the performance of those potential solutions -- reducing their likelihood of outcompeting an existing elite in that new niche. We utilize a technique, we call Pollination, that periodically replaces the controllers of certain solutions with a distilled controller with better generalization across morphologies to reduce fragile brain-body co-adaptation and thus promote MAP-Elites migrations. Pollination increases the success of body mutations and the number of migrations, resulting in better quality-diversity metrics. We believe we develop important insights that could apply to other domains where MAP-Elites is used.

Problem

• Brain-body co-optimization suffers from fragile co-adaptation
• MAP-Elites disrupted by brain-body specialization hindering migrations
• Pollination technique reduces co-adaptation to improve solution transfers

Innovation

• Controller distillation reduces fragile brain-body co-adaptation
• Pollination technique enhances MAP-Elites migrations
• Generalized controllers improve quality-diversity metrics