Strategic Inertial Measurement Unit for Wheel Odometric Encoding Sensor Fusion System in Mobile Autonomous Robots

Localization, a fundamental aspect of robotics, revolves around determining the precise position and orientation of a robot within its environment. The challenging aspect of localization lies in accurately estimating the robot’s position and orientation, which is crucial for effective navigation and interaction with the environment. This vital information is acquired through the integration of sensor data and other systems, allowing the robot to precisely ascertain its whereabouts in the environment. In this study the integration of IMU sensors and wheel encoders to estimate the robot’s 6 degrees of freedom (DOF) ego-motion was examined. A further study was done on the use of an Extended Kalman Filter (EKF) for the fusion of the IMU and Wheel Encoder sensor readings sensor fusion. Both physical and simulated robots were utilized in our experiments, all designed to operate within ROS2 ecosystems. Our findings indicate that the default Gazebo settings, `x`-axis errors averaged around -1.75 with high variability (standard deviations of 1.90), and `y`-axis errors averaged around -0.57 with moderate variability (standard deviations of 0.54). Adding noise exacerbated `x`-axis errors to a mean of -2.15, but filtering reduced this to -1.15. `Y`-axis errors remained stable across datasets with a mean around -1.33. Although filtering enhanced `x`-axis accuracy, variability persisted. These results underscore the impact of noise on precision and the advantages of filtering, highlighting the critical role of sensor accuracy.