Problem
Research questions and friction points this paper is trying to address.
multi-objective optimization
Pareto-optimal policies
robot design
reinforcement learning
co-design
MOCHA: Multi-Objective Co-Design using Hypernetwork Architectures
🤖 AI Summary
This study addresses the challenge of multi-objective co-optimization for robot morphology and control policies by proposing the MOCHA framework. This method introduces a novel single-network multi-objective co-design paradigm based on deep reinforcement learning, incorporating a Multi-objective Design Hypernetwork (MDH) to represent complex Pareto-optimal policy families, combined with an evolutionary search algorithm to efficiently construct the design Pareto front. Experiments conducted on two robot morphologies validate the effectiveness of this framework. Specifically, a single network successfully generates Pareto-optimal design-policy combinations for two to three conflicting objectives and identifies versatile robotic solutions that maximize cumulative performance across all objectives.
📝 Abstract
In this work, we present MOCHA, the first, to our knowledge, reinforcement learning based approach to computing a family of Pareto-optimal policies across the design space of a robot using a single network. Specifically, MOCHA leverages the hypernetwork architecture to learn a network that produces specialized network parameters optimized for a given objective and parameterized robot design; we term this a multi-objective design hypernetwork (MDH). We demonstrate the capabilities of MDHs to represent a complex family of design-dependent strategies on two distinct robot morphologies, each with six design dimensions and across 2-3 objectives. Moreover, we propose an approach for efficiently producing a Design Pareto set using evolutionary search of the learned policy network, generating the optimal design-policy combination for each objective prioritization. Lastly, we provide an efficient method for computing generalist robot designs which achieve the best cumulative performance across the entire set of objectives.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Hypernetwork
Multi-Objective Reinforcement Learning
Pareto Optimization
Co-Design
Evolutionary Search
🔎 Similar Papers
No similar papers found.
