SLAM using 2D lidar
Video(better watching in 1080p): https://v.qq.com/x/page/q0363h0i1ej.html
- Run the main.m
- If you'd like to test the loop closure detection(only detection, no pose graph optimization yet) functionality, dequote the 'Loop Closing' codes in the main.m
- Register the laser points in a probability-grid-map, it will help improve the scan-matching performance.[1]
- Tightly couple the Laser and IMU to improve robustness and efficiency.
- Estimate relative pose between two consecutive keyscans, estimate the relative pose's covariance following the approach in [1]
- Use pose graph optimization to close loops.
- Use branch and bound method to speed up brute force scan matching.
Though it's not so accurate or robust yet, I believe it can have high performance after finishing the tasks in the ToDo list.
[1]W.Hess, D.Kohler, H.Rapp and D.Andor. Real-Time Loop Closure in 2D LIDAR SLAM. ICRA, 2016