Max Robotics — Global Service Robot Directory + US Certification

About

Parijat Dewangan is a robotics researcher whose work centers on humanoid robot control, inverse kinematics, and multi-task reinforcement learning. His most impactful contribution, "A deep reinforcement learning approach for dynamically stable inverse kinematics of humanoid robots" (2017, 58 citations), tackles the critical challenge of maintaining balance during real-time motion planning. By leveraging deep reinforcement learning, Dewangan developed a method to generate joint-space trajectories that ensure dynamic stability—a fundamental requirement for humanoids operating in unpredictable environments. This work addresses a key bottleneck in humanoid robotics: the need for fast, stable inverse kinematics solutions that prevent falls during complex movements. Dewangan also contributed to hardware design with "Design and development of a humanoid with articulated torso" (2016, 5 citations), where he modified the Poppy platform for heavier loads using high-torque MX-64 servos, enhancing balancing capabilities in varied workspaces. His later research, including "DiGrad: Multi-Task Reinforcement Learning with Shared Actions" (2018, 5 citations), explores efficient learning across multiple tasks with shared action spaces, while "Learning Dual Arm Coordinated Reachability Tasks in a Humanoid Robot with Articulated Torso" (2018, 2 citations) proposes faster online planning for bimanual coordination. Collectively, Dewangan’s work advances the frontier of stable, adaptive humanoid control, bridging reinforcement learning and practical robotic design.