@Eric-Katzfey We are conducting outdoor tests for collision prevention using the M0015 stereo camera. The main parameters we adjusted include ae_desired_msv in voxl-camera-server.conf and blur_size, post_median_size, and cost_threshold in voxl-dfs-server.conf. However, we have encountered some issues that have led to subpar performance in collision prevention. The problems are as follows:
- The influence of lighting conditions such as backlighting, glare, etc., can cause the stereo camera to malfunction and make it challenging to find a set of parameters that work well in all situations.
- When the drone stops in front of an obstacle, sometimes it suddenly moves forward a short distance and then stops again. I suspect it may be due to the drone's body offset or lighting conditions causing a momentary failure in obstacle detection by the stereo camera.
- Would higher resolution cameras better capture obstacle features and thus improve the effectiveness of collision prevention?
Regarding the above issues, do you have any suggestions or ideas? Any guidance you can provide would be greatly appreciated. Below are the parameters I'm currently using.
voxl-camera-server.conf
voxl-dfs-server.conf