2D environment mapping and self-position estimation with ultrasonic range sensor array

Abstract

Currently, modern robots use information from a Light Detection and Ranging (LiDAR) module sensor to build a map of the surrounding environment and simultaneously determine its location within the map. The map information is crucial for many tasks, such as path planning and obstacle avoidance. However, the LiDAR Module is expensive compared to other distance sensors, such as ultrasonic sensors. Therefore, this project will use low-cost ultrasonic sensors installed on the two-wheel-drive education-grade robot to build map. Then, the odometer data from the robot’s wheels and distance data from ultrasonic sensors are passed to the Particle Filter (PF) -based SLAM algorithms to precisely specify the robot’s position. The imprecise map created from running the robot in an L-shape map reveals that using inaccurate information from the low-cost sensors and education-grade robot directly affects the quality of the created map. Therefore, morphological image processing is applied to the created map to improve the map quality. As a result, the similarity is increased to approximately 70% compared to the ground truth map. We need to control the robot precisely in different positions to get quality results. Nevertheless, it is hard to do by using educational grade robots. Accordingly, we push the robot by hand in our experiments instead of controlling the motor.