Carlos Valle

Technical University of Munich, Pontifícia Universidade Católica do Rio de Janeiro, Institute of Robotics

Papers

Total Citations

H-Index

About

Carlos Valle is a roboticist whose research lies at the intersection of manipulation, human-robot interaction, and learning-based control. His work addresses fundamental challenges in making robots more adaptive and safe in dynamic environments. Valle’s most cited paper, "Real-time IMU-Based Learning: a Classification of Contact Materials" (2022, 7 citations), introduces a novel sensing approach that uses inertial measurement units to distinguish between deliberate, performance-enhancing collisions and accidental impacts during manipulation—a critical capability for next-generation human-robot collaboration. Earlier, in "Computed-torque control of a simulated bipedal robot with locomotion by reinforcement learning" (2016, 4 citations), he developed a hybrid control framework combining model-based dynamics with reinforcement learning to achieve stable locomotion for an Atlas humanoid in simulation. More recently, his work "Object-Centric Grasping Transferability: Linking Meshes to Postures" (2022, 2 citations) tackles the problem of generalizing human-inspired grasps across different objects by placing the object’s geometry at the center of the learning process. Together, these contributions showcase Valle’s commitment to bridging perception, control, and learning for more capable and intuitive robotic systems.