Tamim Asfour, professor at Karlsruhe Institute of Technology, presents research on robotic manipulation at ICRA 2025, following an introduction by Kevin Lynch of Northwestern University.
Asfour, who has worked on humanoid robots for more than 25 years, describes manipulation as a fundamental challenge for robotics. He argues that changing the world through physical interaction is closely tied to intelligence.
Asfour proposes taxonomies as a way to organize the constraints involved in manipulation and connect perception, planning, learning, and control. His bimanual taxonomy distinguishes independent actions from coordinated ones, including loosely coupled tasks with spatial or timing requirements and tightly coupled tasks that also involve forces.
Asfour describes using motion capture and visual demonstrations to identify task categories and geometric constraints. In a pouring example, a point on the kettle is related to a curve on the cup rim; the extracted constraints converged after eight demonstrations. He says sparse key points can support transfer across different cup shapes and pouring styles.
A taxonomy of 46 whole-body support poses describes contact between a humanoid and its environment; stepping up stairs can be represented by four poses. Asfour says learned pose transitions can support whole-body motion generation. He also cautions that taxonomies can oversimplify force-based, context-dependent tasks and do not handle deformable objects.
In Q&A, Asfour says human movement remains an important source of inspiration, while taxonomy knowledge can guide learning from data. He notes that visual imitation uses vision data, but manipulation also needs force and tactile information. His group has started collecting force data and sees limits in simulation for accurately modeling contact.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.
Tamim Asfour, a humanoid robotics researcher at KIT in Karlsruhe, discusses motion intelligence and versatile robots for household, industrial, and assistive tasks at the Latin American Summer School on Cognitive Robotics.
