I am working on a project as part of ProjX (who recommended I reach out to the Machine Intelligence Community), to make a robotic arm to autonomous connect the charger for a Tesla Model 3 (possibly will work with Y, but I don’t have access to one to measure particulars).
My CAD in progress is shown here.
My plan for computer vision (on Raspberry Pi) was to:
1) classify images on a regular interval (even every 60s is plenty fast) to determine whether the model 3 is present (preferable to discriminate vs other vehicles)
2) use object detection to estimate location of the charge port
3) once charge port is open, estimate location of apriltag placed a known distance from charge port hole.
I have taken 2.120 introduction to robotics and have a cursory understanding of these topics, but only within the context of simplified sample cases and datasets. I have been working through the intro to deep learning course online (6.s191) and some tutorials on pyimagesearch.com, but I am having a hard time understanding how to choose a model type for this effort, if I should just abandon portions of my plan, etc. I have seen tutorials for various models that don't seem to fit my use case, and am a little overwhelmed with all the paths I could take.