Experiments with DeepestLSTMTinyPilonet and Pilotnet on more circuits

less than 1 minute read

More experiments conducted using Deepest LSTM TinyPilotNet and PilotNet on even more circuits.

The lines that cross lap seconds and orientation mae reflect the mean for each metric using the explicit brain.

On the lap seconds metrics, we can see thah both PilotNet and Deepest are faster than the explicit brain for all the circuits. From the lap seconds metrics we can also learn that PilotNet seems to be faster than Deepest LSTM.

Comparing the Orientation MAE, the conclusions are different. The explicit brain is better for all the circuits. For Montmeló the error is much higher for every approach than on any other circuit, which means that this circuit is more difficult. For the rest of the circuits, the explicit brain behaves quite better than the rest of the brains. Considering Deepest and PilotNet, Deepest results are slightly better for all the circuits, even completing Montmeló more times than PilotNet.