I understand that while using the app, its assuming the ego moving and creating an unexpected map.
The behaviour you're describing in the SLAM (Simultaneous Localization and Mapping) Map Builder app in MATLAB typically happens when there's a mismatch between the provided sensor data (scans) and the ego motion information (poses or odometry). Here are some insights and steps to address the issues:
- Ego Vehicle Motion: The SLAM algorithm relies on both sensor data (scans) and ego motion information (poses or odometry) to estimate the map and the vehicle's pose within that map. If you have stationary scans but don't provide any ego motion information, the algorithm may assume motion, leading to unexpected map results. Providing accurate ego motion information is essential, especially when the vehicle is stationary.
- Providing Poses or Odometry: To improve SLAM performance, it's recommended to provide accurate ego motion information (poses or odometry) along with the scans. This helps the algorithm understand the vehicle's motion and can lead to more accurate map estimation. Ensure that the provided ego motion data corresponds to the timestamps of the scans.
- Downsampling Data: Downsampling can help reduce clutter and improve the efficiency of the SLAM algorithm. You can consider downsampling the scans to reduce the number of data points while maintaining essential information. The downsampling factor should be chosen carefully to balance data reduction with map quality. MATLAB provides functions like pcdownsample for downsampling point cloud data.
Refer to this documentation for more information:
