Two robotic arms manipulate a volumetric piece of foam controlling its shape using visual feedback from an overhead 3D camera.
The ability to effectively manipulate deformable objects enables the robotization of crucial tasks in industrial, agricultural, domestic, or medical domains. By combining suitable deformation modeling and visual sensory feedback, we have developed versatile manipulation systems applicable to, e.g., industrial assembly of flexible parts or pruning of plants.
The videos below show examples of our recent work in this area.
Unified approach for tracking and controlling the shape of deformable objects, article in IEEE TRO (TRO24)
Mimicking dynamic deformations using a geometric model, additional results from article in IEEE RAL (RAL22a)
Deforming elastic linear objects with an adaptive control approach, article in Frontiers in Robotics and AI (FRAI22)
Deformation control of soft linear objects using deep reinforcement learning, article in IEEE RAL (RAL24a)
Controlling the deformation of linear objects with a simple offline model, article in IEEE/RSJ IROS 2022 (IROS22)