This question isn't just for teachers, but I wasn't sure which other category with be better.
Which robots, other than Finch, work with Snap! ? I would like to use Snap! to program a robot to move through a maze, including detecting and distinguishing between two different kinds of objects (e.g., metal cylinder, wooden cube) that are obstacles in the path. Finch does not distinguish objected detected; perhaps no robot on the market does. Even so, I would like to eventually have a robot that uses a block language and can distinguish objected encountered.
The ultimate goal is a form of robot chess with two robots on a grid moving at the same time, attempting to be the first to either exit their door out of the grid or contact the other robot and so "capture" it as in chess.
There's a list of robot support packages here: https://snap.berkeley.edu/extensions
But the Finch (and its big brother the Hummingbird kit) has the big advantage that its support software is maintained by the manufacturer. Most of the others aren't maintained at all, and so there's no guarantee about which ones still work.
Bh and snapenilk, do you know whether any of these physical robots can distinguish between objects detected such as mentioned (metal cylinder, wood cube)?
The ones that are kits in which you can place components where you want could, I guess, shine a light forward and measure how much reflection they get back. But I've never tried it.
I think you need to approach this ambitious project in stages
I'd try and leverage existing object recognition systems (Raspberry Pi and with camera comes to mind) and get them to be able to identify difference between a wall, an exit or another object
Controlling the movement of a robot itself can be done many ways
One way is to run a python webserver on a Pi to talk to Snap!
The python webserver talks to another python program that does the motor controls (and ultrasonic sensors and the like)
As I said, the hard part will be the AI object stuff