THE ACTIVE RECOVERY OF 3D MOTION TRAJECTORIES AND THEIR USE IN PREDICTION

This paper describes the theory and real-time implementation using an active camera platform of a method of planar trajectory recovery, and of the use of those trajectories to facilitate prediction over delays in the visual feedback loop. Image-based position and velocity demands for tracking are generated by detecting and segmenting optical flow w ithin a central region of the image, and a projective construct is use d to map the camera platform's joint angles into a Euclidean coordinat e system within a plane, typically the ground plane, in the scene. A s et of extended Kalman filters with different dynamics is implemented t o analyze the trajectories, and these compete to provide the best desc ription of the motion within an interacting multiple model. Prediction from the optimum motion model is used within the visual feedback loop to overcome visual latency. It is demonstrated that prediction from t he 3D planar description gives better tracking performance than predic tion based on a filtered description of observer-based 2D motion traje ctories.