Evaluation on IMU and odometry sensor fusion for a Turtlebot robot using AMCL on ROS framework.

Abstract

When a robot’s coordinate system and the environment map have a correspondence with localization algorithms, problems arise related to the accurate and direct measurement of the robot’s position, which is directly linked to sensor measurements that often contain errors, noise, and influencing factors. In this article, an experiment will be conducted to demonstrate error propagation and the attempt at correction using the mentioned algorithms. This study will involve comparing simulation and odometry results from a real robot through two different trajectories. Correction techniques were used through the ROS framework to assess the robot’s position using the Extended Kalman Filter (EKF), which performs a fusion of data from Odometry sensors, Inertial Measurement Unit (IMU), and the Adaptive Monte Carlo Localization (AMCL) algorithm on a Turtlebot3 in an indoor environment. Additionally, two navigation tests were conducted, one using the map generated by simultaneous localization and mapping (SLAM) and another using a computer-aided design (CAD) map of the environment where the robot is navigating. In this sense, the tests provide us with a comparison of sensor localization and localization algorithms in CAD and SLAM maps.