Visual Aided GPS Navigation for Autonomous Mobile Robots

Navigation based on GPS data has been the most commonly used methodology for the autonomous run of a mobile robot in indoor and outdoor environments. However, the reading of the GPS receivers fluctuates over a considerable range esp. in countries like India where there is a dearth of GPS signals and locally available correction table. So the commonly received value by GPS receiver can be accurate over a range of 10-15mts which is inadequate if the test results require accuracy. This paper introduces a technique for the development of a visual aided GPS navigation system for a mobile robot in which we have predefined visual landmarks for the various important landmarks which robot has to visit. In our approach, we have mounted a stereovision camera on the robot platform for image acquisition, real time object recognition, detection and local features extraction from images using Scale-Invariant Feature Transform (SIFT).