Sample code for the conference paper "Landmark management for the application of Radar SLAM" (ArXiv:2209.07199)
Author: Shuai SUN, Chris GILLIAM, Beth JELFS
Email: shuai.sun@dlmu.edu.cn
The sample code provides an example of landmark (vehicle) management for radar SLAM.
Trajectory of the mobile platform is fixed, and a motion noise is added as pertubation
There are 2 different radar models which can be set using the parameter sensor.model in the loadParameters script:
- Ideal this model directly generates landmark detections by drawing randomly from within the landmark region
This is the model used in the paper - PhasedArray this models a phased array with 4 arrays each covering 90 degrees
Detections are obtained using a CFAR algorithm
To reproduce the results obtained in the paper, the ideal model should be used with the following parameters (Table I in the paper)
| Symbol in Paper | Name in Code | Value |
|---|---|---|
| Rmax | sensor.max_range | 20 |
| Rk | sensor.R_radar | [0.5^2 0; 0 (pi/180)^2] |
| Qk | sensor.Q_process | [1.5e-3, 0 0; 0 1.5e-3 0; 0 0 5e-5] |
| Uk | sensor.U_odometer | [0.022^2, 0; 0 (0.008 * pi/180)^2] |
| α | thresholds.alpha | 500 |
| β | thresholds.beta | 20 |
| γc | thresholds.cluster_radius | 2.5 |
| γs | thresholds.sifting_radius | 3 |
| γa | thresholds.association_radius | 3.5 |
| γm | thresholds.merge_radius | 1.5 |
| Nc1 | thresholds.point_threshold | 6 |
| Nc2 | thresholds.min_cluster_points | 2 |
| M (landmark initialization) | MN_logic.M_association | 5 |
| N (landmark initialization) | MN_logic.N_association | 3 |
| M (landmark removal) | MN_logic.M_deleting | 10 |
| N (landmark removal) | MN_logic.N_deleting | 2 |