@Maximellerbach - can you have a look at this?

@DocGarbanzo the type of self.info['lidar'] should be a np.array, I'm not sure of the expected behavior when the array is empty ? Are we checking somewhere if the array of point is empty / matches the expected shape the model needs ?

@DocGarbanzo the type of self.info['lidar'] should be a np.array, I'm not sure of the expected behavior when the array is empty ? Are we checking somewhere if the array of point is empty / matches the expected shape the model needs ?

Currently the lidar data is not used in any models. For now it's just being recorded into the tub, as it would be returned in the run_threaded method. Here, we would need a native python list, but not an np.array.

Currently the lidar data is not used in any models. For now it's just being recorded into the tub, as it would be returned in the run_threaded method. Here, we would need a native python list, but not an np.array.

Maybe I'm missing something, why should it be a python list ? I am currently returning from GymDonkeycar a numpy array, if you confirm it should be a list then I will also change it on this side

I think the older drivers return an np.array() and the newer drivers return a python list. I would go with the newer drivers. This is what the new driver puts out:

Ok ! I will then align with those.
What does the time and scan values look like ?

So I think we can merge this @DocGarbanzo