SIGMA has coordinator has a full overview of the activities performed in these two years by SOFTMANBOT project. For Robotic perception, a novel pipeline named “ROBUSfT” has been proposed that generates the 3D shape of deforming objects from images coming from a monocular camera. Furthermore, a software architecture is under development that connects all project parts for all partners. For AI driven contact based task control, a dynamic evaluation method of deformable object grasping has been validated experimentally on the dual-arm robotic platform. As well as, a unilateral teleoperation system, both for grasping and manipulation, has been presented to be used for retrieving data from human demonstrations. Finally, a novel approach to manipulate deformable objects has been proposed and it has been validated on bi-arm manipulation experiments at SIGMA’s mock-up with different objects including a tire tread. These experiments will be tested at our industrial partner Michelin.