Max Robotics — Global Service Robot Directory + US Certification

About

Somil Bansal is a robotics and control researcher whose work sits at the intersection of optimal control, machine learning, and safety-critical systems. His research primarily addresses how autonomous robots can navigate complex, partially observable environments while maintaining rigorous safety guarantees — a challenge of growing importance as robots move into human-centered spaces. Bansal's most influential contribution, "Combining Optimal Control and Learning for Visual Navigation in Novel Environments" (2019, 96 citations), exemplifies his signature approach: marrying model-based control with learning-based perception to enable robust robot navigation in previously unseen settings. Complementing this, his work on goal-driven dynamics learning via Bayesian optimization (2017, 76 citations) introduced a task-specific framework for learning robot dynamics in poorly understood environments, reducing the burden of exhaustive full-system modeling. A recurring theme across his body of work is safety assurance in dynamic, human-occupied environments. His research on Hamilton-Jacobi reachability analysis, human motion prediction, and confidence-based safety updates reflects a deep commitment to deploying robots reliably near people. Additional contributions spanning legged robot stability, vision-based controller failure discovery, and imitation learning further demonstrate his broad technical range. Collectively, Bansal's research provides foundational tools for building autonomous systems that are both intelligent and provably safe.