Kalman Filter Enhanced Tracking Controller for Mobile Robots with Bounded Accelerations

This paper presents an extension of an existing tracking controller that is a neural dynamics based tracking controller that uses a gated dipole model [12]. The existing approach that is extended succeeds in eliminating speeds jumps, handling discrete paths, removes the perfect velocity tracking assumption and is computationally efficient. The extension that is made to the work in [12] replaces the path integration module that is used to track the robot´s current position with a more reliable extended Kalman filter (EKF) position tracker and the addition of an orientation sensor to deal with an environment that includes noise. The improved controller shows measurable improvement over the existing controller in situations where non-constant reference velocities are used.