A Study of Using a moving average filter for reducing noise in distance data from ultrasonic sensors used in obstacle aviodance of autonomous ground vehicles

Abstract

This research article investigates a method to mitigate cross-talk interference between two identical ultrasonic sensors used for obstacle detection in front of an Automated Guided Vehicle (AGV). The proposed approach involves alternating the operation of each sensor at 50 ms intervals, combined with the application of a moving average filter using 5 and 10 data points. The study is divided into 1) The experiment with both ultrasonic sensors simultaneously measuring distance to a wall without using the alternating control program showed that the sensors exhibited measurement errors due to interference occurring every 18–19 seconds. The moving average filter helped smoothly the distance data in the time domain but was not sufficiently effective. 2) Both ultrasonic sensors are set to detect objects immediately. The sensors are mounted on a cart and move along a rail that moves across the front of a target object located at distances ranging from 40,70 and 100 centimeters. The result that shown in raw data of distance measurements occurred greatest mean deviation of 100 centimeters at 48.06 centimeters. Applying a moving average filter with 5 and 10 can reduced fluctuations by 80.25% and 88.24%, respectively. 3) The AGV robot automatically stops when the ultrasonic sensors detect object at distances of 40, 70, and 100 centimeters. The resulting is shown moving average filter coefficients of 5 and 10 are stop away from 22.7 and 37.1 centimeters, respectively. In case of 40 centimeters and defined filter coefficients 10, the robot cannot process the stop command and collides with an object