Final Report Summary - FLEXBOT (Flexible object manipulation based on statistical learning and topological representations) We use hands to tie shoes, write, wash dishes, play piano: these tasks involve close and complex interaction with objects and we can execute them even if one of our hand/arm/fingers is injured or not functioning properly. Our dexterity and sensory capabilities have developed through a long process of evolution - however, for the above tasks, coupling the sensing and dexterity with efficient representations, planning and execution is necessary. Compared to humans or primates, the sensing and dexterity of today’s robotic hands and medical prostheses is extremely limited. The latter commonly have only a single degree of freedom allowing them to grasp and manipulate a limited set of objects. Replicating the effectiveness and flexibility of human hands requires a fundamental rethinking of how to exploit the available mechanical dexterity. A vision for the future are systems that perform complex tasks safely and robustly in interaction withhumans and the environment. This requires integration of sensory data in order to estimate the state of the environment, understand the requirements of a specific task, and reason on its own motor ability for planning and acting in uncertain, dynamic environments. These systems cannot be fully preprogrammed for all the tasks they are supposed to execute. One approach is to equip them with the capability to learn from humans: in robotics, this approach is known as Learning by Imitation. As one of the most ubiquitous forms of learning in nature, imitation presents an important research problem in robotics, AI, machine learning, and neuroscience. Though proved as an effective approach, the existing methods train the robot of how to reach for an object but in-hand manipulation is usually based on advanced sensory control. Interaction with objects in complex scenes requires a huge amount of computation for collision avoidance and effective task planning. The outcomes of the project provide a better insight into the complexity of the above problems. We addressed problems such as optimal grasp planning, caging and motion planning as well as task representation.
