Max Robotics — Global Service Robot Directory + US Certification

About

Yuxiang Yang is a robotics and machine learning researcher whose work sits at the intersection of reinforcement learning, evolutionary computation, and adaptive robot control. His most recognized contribution, "Rapidly Adaptable Legged Robots via Evolutionary Meta-Learning" (2020), has garnered over 60 citations and presents a novel meta-learning framework enabling legged robots to swiftly adjust to dynamic changes in their environment — a critical capability for real-world autonomous operation. Unlike traditional gradient-based approaches, his method leverages evolutionary strategies to achieve robust, efficient adaptation. This theme of combining evolutionary computation with modern deep learning runs throughout his research portfolio. In his 2019 work on provably robust blackbox optimization, Yang strengthens the theoretical foundations of derivative-free optimization for reinforcement learning, addressing known weaknesses in evolutionary strategy methods. His 2021 paper, ES-ENAS, further demonstrates his creativity by seamlessly merging evolutionary strategies with neural architecture search to automate policy design for RL agents at scale. Together, these contributions reflect a researcher dedicated to making robot learning more adaptable, theoretically grounded, and practically deployable — offering valuable tools for the next generation of intelligent autonomous systems.