Commute Guardian

[Luxonis-Brandon]

Start with the why:

It's time. We've implemented all the basis functions of the platform (https://github.com/orgs/luxonis/projects/2), all of which were guided by the North Star of keeping people who ride bikes safe (here).

Namely, we now have all the pieces that go into building this below, in hardware, firmware, and software:

• Onboard depth sensing to 30+ meters
• Wide FOV cameras for detecting more side-impact situations
• Lossless zoom at 12MP, for recording and recognizing license places quite far away
• Retraining of vehicle object detector that runs at high FPS (greater than 30FPS) and long-distance (greater than 30 meters)
• Feature tracking, edge filtering, semantic segmentation that can all run in parallel - for vehicle-edge tracking in physical space
• Integrated host support (running YOCTO Linux)
• High-bandwidth WiFi + BT support (for streaming to smartphone)

And likely a bunch of others.

Move to the how:

For the first version:

• Leverage our OAK-SoM-Pro (here) with MA2095 (Keem Bay) populated on a baseboard (i.e. go with a baseboard + SoM approach).
• Use the ~150-degree OV9282 modules from OAK-D W (#152). Orient such that the widest field of view is parallel to the horizon.
• Use the ~120-degree IMX378 (fixed focus) from the variant of OAK-D W that we're preparing (not yet documented). Orient such that the widest field of view is parallel to the horizon.
• BT and WiFi connected to MA2095/OAK-SoM-Pro directly.
• Plan on e-bike deployment only, with 5V and at least 1A input as a requisite.
• Open Source the whole thing.

Move to the what:

Make the first version of Commute Guardian. Probably something that looks like this:

• IP67 sealed