Integrating Visual and Encoder Data for Enhanced Navigation in Mecanum-Wheeled Robots via Extended Kalman Filter

This paper presents a methodology for fusing data from the IMU data acquired from the Intel RealSense 415i camera and the wheel encoder of a mecanum robot to estimate its velocity and acceleration. The proposed approach utilizes an Extended Kalman Filter (EKF) to integrate these information to enhance its navigation capabilities. the proposed approach also utilizes the capabilities of the companion computer onboard the robot to process the visual data for Simultaneous Localization and Mapping (SLAM). The paper discusses the implementation details, performance evaluation, and experimental validation using a real robot. From the results, the positional error using EKF was about 56 mm, exhibiting superior performance compared to the raw sensor data streams.