I’m working on detecting obstacles the ping 360 captures, Do you have an algorithm recommendations, or any example of how obstacle detection could be implemented. I’m using python. And I’ve got another doubt, if I were to implement a sector scan, and If I had to do a 180 deg sector scan, with start angle as 270 and end angle as 90, It seems that auto_transmit wouldn’t work for that.
Hi @issac5693,
Cool! There have been some previous discussions around that which may be of interest 
The simplest approach is just some kind of thresholding algorithm, where you consider each profile as an independent obstacle detector. More advanced approaches involve correlating raw detections between profiles (e.g. to get some estimates around size and shape of obstacles), and potentially tracking obstacles relative to the vehicle over time, but both of those require compensation for vehicle movements during the scanning period.
Where you focus on performance improvements depend on what kind of information is actually valuable to your use-case, trading off computation resources, hardware costs, and spatial and temporal resolution.
Assuming you mean going through the zero angle, yes, I believe that is a limitation of the current firmware’s auto_transmit support. There is a minimal Issue to that effect, but it hasn’t so far been determined to be a priority.
Is such an algorithm implemented? where could I find the implementation of such an algorithm. could you provide any information on the algorithms that I could use?
Blue Robotics does not have implementations of such an approach, if that is what you’re asking.
A brief online search yielded this research paper.
I imagine there are also several generally relevant examples in the radar / lidar space.
Thanks a bunch Eliot!