Contact State Estimation for Multi-Finger Robot Hands Using Particle Filters

A method, using a particle filter, has been created for identifying the location, orientation, and shape of an object that a robot hand touches. An appropriate motion model has been defined that characterizes the motion of the robot hand as it moves relative to an object. The measurement model also estimates the likelihood of an observation of contact position, velocity, and tactile sensor information given handobject states, and is approximated analytically based on a geometric model, or on a corpus of training data. In either case, the measurement model distribution is encoded as a Gaussian, or using radial basis functions.