Robot Manipulation and Learning

Robot manipulation and learning sits at the intersection of control theory and machine intelligence, concerned with how robotic systems can reliably pick up, reorient, and interact with physical objects in environments that were not precisely engineered for them. Core challenges include estimating where an object is and how it is oriented in three-dimensional space, then translating that information into smooth, force-aware motion that can adapt when something goes wrong. Learning from human demonstration—using techniques like dynamical movement primitives or deep neural networks trained on sensor data—has become a central strategy for acquiring manipulation skills without hand-coding every contingency. Open questions include how to make learned behaviors generalize robustly across unfamiliar objects and clutter, and how to ensure that robots working alongside people can detect and respond to human intent quickly enough to remain genuinely safe.

Works

61,554

Total citations

685,502

Keywords

Robot LearningGraspingDeep LearningHuman-Robot CollaborationObject Pose EstimationDynamical Movement Primitives