Research on Positioning Accuracy of Mobile Robot in Indoor Environment Based on Improved RTABMAP Algorithm

Area of Science:

• Robotics
• Computer Vision
• Simultaneous Localization and Mapping (SLAM)

Background:

• Visual SLAM is crucial for mobile robot positioning where GNSS fails.
• Indoor robot navigation suffers from cumulative position errors due to environmental factors.
• Existing methods like RTABMAP can experience decreasing accuracy over time.

Purpose of the Study:

• To improve the accuracy and reduce cumulative errors in Visual SLAM for mobile robots.
• To enhance the RTABMAP method by integrating Inertial Measurement Unit (IMU) data.
• To evaluate the performance of the proposed RTABMAP-VIWO method against existing SLAM techniques.

Main Methods:

• Proposed RTABMAP-VIWO method based on the RTABMAP framework.
• Utilized an Extended Kalman Filter (EKF) to fuse wheel odometry and IMU attitude estimates.
• Provided new prediction values to mitigate local cumulative errors.

Main Results:

• RTABMAP-VIWO reduced Root-Mean-Square Error (RMSE) by 48.1% compared to RTABMAP on public datasets.
• Achieved at least a 29.4% reduction in RMSE in real-world indoor experiments.
• Demonstrated significant improvements in trajectory and posture error reduction.

Conclusions:

• The RTABMAP-VIWO method effectively reduces cumulative errors in indoor mobile robot localization.
• Incorporating IMU data into RTABMAP significantly enhances positioning accuracy.
• The proposed method offers a feasible and more accurate solution for VSLAM in challenging indoor environments.