INTEGRATED MULTISENSORY ROBOTIC HAND SYSTEM FOR DEFORMABLE OBJECT MANIPULATION

摘要

Designing a dexterous robotic hand able to interactrnintelligently with deformable objects constitutes arnchallenging area of research where many issues are yet tornbe solved. The complexity of such interactions requiresrnthe assistance of intelligent multisensory robotic systemsrnthat combine measurements collected from differentrnsensors in order to accurately plan for the forces to bernapplied on the deformable object. This paper presents therndevelopment of a real-time multisensory robotic handrnplatform that incorporates live measurements of itsrninternal position, velocity and force parameters along withrndata from external tactile sensors and a stereoscopicrnvision device. The resulting prototype of the integratedrnmultisensory system is validated experimentally by therncomputation of deformable object models in which thernmeasurements are merged. A formal dynamic model isrndiscussed and a neural network representation model isrnpresented. The results demonstrate the performance andrnsuitability of the multisensory platform for therndevelopment of enhanced robotic hand capabilities whenrnmanipulating deformable objects.