Zhehui Huang

Southern California University for Professional Studies, University of Southern California

Papers

Total Citations

H-Index

About

Zhehui Huang is a researcher at the intersection of reinforcement learning, robotics, and multi-agent systems, with a focus on scalable and efficient learning frameworks for autonomous control. His most influential work, "Sample Factory" (2020), introduced a high-throughput asynchronous reinforcement learning framework capable of achieving 100,000 frames per second on a single machine, dramatically reducing the computational cost of large-scale RL experiments without relying on expensive distributed infrastructure — earning 16 citations and establishing him as a contributor to accessible, high-performance RL tooling. Building on this foundation, Huang has directed significant effort toward quadrotor swarm intelligence. His 2024 paper on end-to-end deep reinforcement learning for quadrotor swarm collision avoidance and navigation (13 citations) demonstrated deployable multi-agent controllers capable of real-time execution, while "QuadSwarm" (2023) provided the community with a modular, parallelizable simulator to support such research. More recently, his exploration of large language models for robot routing problems signals a growing interest in bridging classical combinatorial optimization with modern AI. Across his work, Huang consistently pursues practical, deployable solutions to complex multi-robot coordination challenges, making his research valuable to both academic and applied robotics communities.