Sonar sensor data processing based on optical flow in robot navigation

Abstract

The application of optical flow techniques on artificial image sequences in computing relative motions of obstacles is studied as a data processing algorithm for ultrasonic sensors. These sensors in robot systems are widely used in calculating the distance of obstacles. However, owing to the angular uncertainty of sensor characteristics, limitations exist on the positioning of an object's location, making it difficult to obtain accurate velocity information. Collision avoidance in dynamic environments requires information on relative motion, such as distance and velocity. In this paper, a technique for image processing on a range of ultrasonic sensor data is suggested in order to obtain the relative motion of front objects. The proposed technique is conducted using a two-wheel differential drive robot for problem cases related to indoor navigation. Digital filtering is also conducted to smooth the fluctuations, after which relative velocity is calculated using the optical flow technique. Experimental results demonstrate the possible application of the proposed algorithm for robot navigation in dynamic environments. This technique will increase the intelligence of robot systems, allowing for effective sensing in various environmental conditions.