Co-RaL: Complementary Radar-Leg Odometry with 4-DoF Optimization and Rolling Contact

Sangwoo Jung,Wooseong Yang,Ayoung Kim

2024-07-10

Abstract:Robust and accurate localization in challenging environments is becoming crucial for SLAM. In this paper, we propose a unique sensor configuration for precise and robust odometry by integrating chip radar and a legged robot. Specifically, we introduce a tightly coupled radar-leg odometry algorithm for complementary drift correction. Adopting the 4-DoF optimization and decoupled RANSAC to mmWave chip radar significantly enhances radar odometry beyond the existing method, especially z-directional even when using a single radar. For the leg odometry, we employ rolling contact modeling-aided forward kinematics, accommodating scenarios with the potential possibility of contact drift and radar failure. We evaluate our method by comparing it with other chip radar odometry algorithms using real-world datasets with diverse environments while the datasets will be released for the robotics community. <a class="link-external link-https" href="https://github.com/SangwooJung98/Co-RaL-Dataset" rel="external noopener nofollow">this https URL</a>

Robotics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to achieve accurate and robust localization in challenging environments, especially in extreme environments (such as fire scenes, mountain trails, and stairs, etc.), where traditional sensors (such as cameras and LiDAR) are vulnerable to visual degradation and feature degradation, leading to potential sensor failures. To this end, the author proposes a unique sensor configuration combining a chip - based radar and a legged robot to achieve accurate and robust odometry.

Specifically, this paper mainly addresses the following problems:

1. **Vertical drift problem of radar odometry**: Since a single - chip radar has low accuracy in estimating the height direction, especially when there are large dynamic objects, it is prone to vertical drift. By introducing the 4 - degree - of - freedom (4 - DoF) optimization and decoupled RANSAC algorithm, the author significantly improves the accuracy of radar odometry in the z - axis direction.

2. **Contact drift problem of legged robots**: When a legged robot walks on complex terrain, it may cause contact drift due to contact sliding or vibration. By introducing forward kinematics assisted by rolling - contact modeling, the author reduces the impact of this drift on the accuracy of odometry.

3. **Multi - sensor fusion problem**: In order to improve the robustness and accuracy of the overall system, the author tightly couples the radar and legged - robot odometry to achieve complementary drift correction. Through this method, even if one sensor fails, the other sensor can still provide reliable odometry information.

### Specific contributions

1. **Significantly improve the accuracy in the z - axis direction**: For the first time, the author applies the 4 - DoF optimization scheme to the radar and introduces decoupled RANSAC for outlier removal in the z - axis direction, thereby significantly improving the odometry accuracy of a single - chip radar in the z - axis direction.

2. **Legged - robot odometry with rolling - contact modeling**: The author considers the rolling motion of the contact surface of the legged robot and achieves a more accurate velocity measurement model by incorporating the rolling motion into the body - velocity derivation.

3. **Real - world evaluation**: Using a real - world radar - legged - robot data set, the author comprehensively evaluates the proposed method and compares it with existing algorithms. The results show that this method exhibits significantly improved accuracy and robustness in z - axis translation. The author will also release these data sets to promote further research in this field.

### Conclusion

By combining a chip - based radar and a legged robot and introducing 4 - DoF optimization and rolling - contact modeling, this paper successfully solves the difficult problem of achieving accurate and robust odometry in complex environments. This provides new ideas and technical means for future SLAM research.