Max Robotics — Global Service Robot Directory + US Certification

About

Nitish Sontakke is a robotics researcher specializing in bridging the sim-to-real gap for legged robots, with a particular focus on quadrupedal locomotion and novel soft robotic platforms. His work addresses the fundamental challenge of transferring policies trained in simulation to real-world hardware, developing methods that combine deep reinforcement learning with domain randomization and system identification. Sontakke’s most impactful contribution is his work on Buoyancy Assisted Lightweight Legged Unit (BALLU) robots, where he pioneered residual physics learning for sim-to-real transfer on these uniquely sensitive, intrinsically safe platforms—a paper that has garnered 9 citations since 2023. He has also advanced quadrupedal locomotion through scalable motion imitation, enabling a single policy to produce diverse behaviors, and developed PM-FSM, a hybrid approach that modulates finite state machines with learned policies for robust locomotion. Most recently, his work on BayRnTune introduces adaptive Bayesian domain randomization with strategic fine-tuning, further refining the sim-to-real pipeline. With a growing citation footprint and a focus on making robots safer and more adaptable, Sontakke is establishing himself as a rising voice in learning-based robotics.