Max Robotics — Global Service Robot Directory + US Certification

About

Yinghao Cai is a robotics researcher whose work spans reinforcement learning, teleoperation, visuotactile sensing, and robot skill acquisition. His research addresses fundamental challenges in enabling robots to learn, adapt, and operate effectively in complex real-world environments. Among his most influential contributions is a shared control teleoperation framework that leverages learning from demonstrations to reduce operator workload during complex remote tasks (35 citations). His work on sim-to-real transfer for deep reinforcement learning offers practical pathways for training robot control policies without the prohibitive costs of purely real-world data collection (29 citations). Cai has also made significant strides in exploration efficiency within sparse-reward reinforcement learning, developing curiosity-driven methods including the ACDER framework (22 citations) that meaningfully accelerate robotic manipulation learning. Notably, his GelStereo 2.0 visuotactile sensor (23 citations) advances high-resolution contact geometry sensing, while complementary learning-based force/torque estimation work further extends tactile intelligence for dexterous manipulation. His meta-learning contributions enable zero-trial robot skill adaptation to novel objects, addressing critical data efficiency challenges. With over 180 cumulative citations across diverse robotics subfields, Cai's research collectively pushes toward robots that learn more efficiently, perceive more richly, and adapt more robustly to real-world demands.