Max Robotics — Global Service Robot Directory + US Certification

About

Aleksei Petrenko is a robotics and artificial intelligence researcher whose work sits at the intersection of deep reinforcement learning, sim-to-real transfer, and high-performance computing for robotic control. He is best known for his contributions to dexterous robotic manipulation, particularly through the **DeXtreme** project, which demonstrated that agile, multi-fingered in-hand manipulation policies trained entirely in simulation can be successfully transferred to real-world robotic systems — a landmark achievement that has garnered over 88 citations since 2023. His follow-up work, **DexPBT**, extended these capabilities to one- and two-armed systems using Population Based Training, further pushing the boundaries of what simulated dexterous agents can achieve. A recurring theme in Petrenko's research is making large-scale reinforcement learning computationally accessible. His **Sample Factory** framework enabled egocentric 3D control from pixels at an astonishing 100,000 frames per second using asynchronous RL, reducing the infrastructure burden typically associated with such experiments. He has also contributed to multi-agent quadrotor simulation through **QuadSwarm** and explored diversity-driven learning via Quality Diversity RL. Collectively, his portfolio reflects a researcher deeply committed to bridging the gap between scalable simulation and real-world robotic intelligence.