I would like to propose that we can still play with the shadow. As I like it better if the wooden robot is separated from the shadow physically, but still somehow able to communicate; what I mean here is that I suggest putting one robot out of the duo in the back of the screen, (with back light) and this is the robot that audience will be able to see the shadow, and the other robot from the duo is in front of the screen, and is reacting the shadow.

Suppose that we have four pairs of robots, then each time only the robot in the light is moving. So, it means that either we have 4 lights, one for each group, and then we code the lights to be on or off at a given time, or use one light that can focus on different spot, by rotating.

Maybe we can use the light sensor, on the back robots, as the sensor detects the light, the script starts to run, and when there is no light, nothing moves. The moment the back one moves, the corresponding front one should start reacting to it.

(Excuse my drawing)

Further thoughts: Maybe we can also add some props made of feathers or other material, which changes dynamically with time in the back of the screen, and let the robot in the front ‘react’ to it.