It is evident that service robotics has the potential to improve people's quality of life and it holds the key to a number of unmet applications related to health care and rehabilitation. According to the prediction of International Federation of Robotics, the global market for intelligent service robots is forecast to reach 24.3 billion USD worldwide by 2010. A multi-fingered robotic hand is the most complex and dexterous robotic system, whose development represents frontiers in service robotics research. Recent innovations in motor technology and robotics have achieved impressive results in the hardware of robotic hands such as Robonaut hand. However, the manipulation systems of robotic hands are hardcoded to handle specific objects in specific ways, which significantly limits their transfer to a range of different situations and applications. The control and optimisation problems involved in robot hand manipulation are very difficult to solve in mathematical terms, however humans solve their hand manipulation related tasks easily using skill and experience. Object manipulation algorithms are required to meet the market requirement that robot hand systems should have human-like manipulation capabilities and be independent of robot hand hardware. Hence, the main challenge that researchers now face is how to enable robot hands to use what can be learned from human hands, to manipulate objects, with the same degree of skill and delicacy as human hands. The proposed work aims to investigate artificial intelligence (AI) methodologies and practical solutions which will allow robotic hands to automatically adapt to human environments and thus to enable them to autonomously perform useful manipulation tasks involved in daily living, pontentially for health care and rehabilitation applications. The investigation will focus on the following areas. 1) To generate a series of responsive human-like finger gaits for a robotic hand given an object to manipulate. This will have the capability to iteratively build a knowledge base representing the features of human hand manipulation behaviour and to efficiently provide corresponding robot hand gaits and manipulation strategies for a given manipulation task in a human environment.2) To develop feasible friction models for the interaction of objects and a robot/human hand. This will enable the application of existing mathematical research findings in multifingered robot manipulation to realworld applications in human environments and will integrate related methods in engineering and AI domains. 3) To develop an AI-based control architecture to ensure robust object manipulation of multifingered robots in terms of manipulation feasibility and efficiency. This will allow robot hands to perform stable human-like object grasping and manipulation and will also provide an open architecture which has the potential to introduce human brain (EEG/MRI signals) and human muscles (EMG signals) information into robotic hand systems.4) To validate the proposed algorithms by implementing these into a set of defined scenarios with a set of simulated multifingered robot hands and three different types of physical robot hands.
|Effective start/end date||23/02/10 → 14/09/13|
- Engineering and Physical Sciences Research Council: £295,150.00
- Electrical Engineering
- Systems engineering
- Control Engineering
- Robotics & Autonomy