The proposed research is directed at developing advanced neural schemes for object-oriented, adaptive, reaching, grasping and manipulation in robotics, with increased stability and robustness. They will be oriented to robotic service applications, and especially for helping people with motor disabilities.
The scientific objectives of the project are:
- Transfer of fundamental biological principles from cognitive-motor behavior to anthropomorphic robotics;
- Construction of a high level, hardware independent, adaptive and modular library of elementary hand gestures;
- New neural schemes that mimic the known cooperation between cerebral cortex and cerebellum/basal ganglia during cognitive motor behaviour in manipulation;
- Specifications for a subsequent industrial research project to be carried out by the endorsers.
The technical objectives of the project are:
- Biological neural networks capable of construct grasp plans in the face of both functional and object specific constraints;
- A language of opposition space and virtual finger, oriented to construct a sequence learning of self adaptive elementary hand gestures with total independence of application and environment;
- Adaptive neural controllers which, tested within laboratory environment using computer simulation, will be capable of generalising their skills in the scaling-up process to different robotics hands. The present project will transfer human planning processes to robotics, while incorporating experimental results from behavioural, anatomical, and neurophysiological studies. A well-defined robust set of mathematical models, elementary library of gestures, behavioural modules, and a selection mechanism over them will produce a considerable improvement in the building of flexible and robust control systems for robot hands. Moreover, it will imply a reduction of the time and efforts necessary in the scaling-up from laboratory to service robotics and a reduction of danger where robots must interact withhumans. The generic neural models for human-like, flexible, adaptive manipulation are not limited to the robotic service industry. The results of the present project are applicable in the whole area of manufacturing where compliance is desirable to interact with the changing environment. BE97- 4099
Funding SchemeCSC - Cost-sharing contracts
LL13 9UZ Wrexham
70100 Tel Aviv