The dynamic vision sensor (DVS) is a silicon retina prototype that records only scene contrast changes in the form of stream of events, thus naturally excluding the redundant background absolute gray levels. In this context, numerous high speed asynchronous event based vision algorithms have been developed and their advantages over frame based processing methods have been compared. In haptic feedback teleoperated micromanipulation, vision is a sound candidate for force estimation if the position-force model is well established. The sampling frequency, however, needs to attain 1KHz to allow a transparent and reliable tactile sensation and to ensure system stability. The event based vision has thus been applied to provide the needed force feedback for two micromanipulation applications: Haptic feedback teleoperation of optical tweezers; Haptic virtual assistance in microgripper based micromanipulation. The results show that the haptic frequency requirement of 1KHz has successfully been achieved. For the first application, high speed particle position detection algorithms have been developed and validated. A three-dimensional haptic feedback system capable of manipulating multiple-trap optical tweezers has been realized. In the second application, a novel event based shape registration algorithm capable of tracking arbitrary form object has been developed to track a piezoelectric microgripper. The stability of the system has been significantly enhanced to assist operators in performing complex micromanipulation tasks.
