Lyapunov based steering control for visual homing of a mobile robot

Abstract

Visual homing refers to navigating a robot from an unknown position towards a predefined home position using visual information. The approach in this paper is based on image correlation extracted from the panoramic images at current and home positions. Our contribution is in designing a Lyapunov function based control-law for homing to a predefined place using the information obtained from matching the panoramic images without estimating homing vector or the bearing information of the current position with respect to the home position. Matching panoramic images render the coarse idea about the change in distance value with respect to the home position. The proposed control-law depends on the output of a signum function. The input to the signum function is the derivative of change in distance function and matching panoramic images provides this information. The performance of the control-law is investigated using simulations with real image data. The proposed control law is in general applicable to applications where aim is to reach the extreme (minima or maxima) value of the sensor information. Applicability of the proposed control-law to different autonomous applications would be investigated in future.