What method are you using for rewarding for following the path? I’m doing something like the below, but it hasn’t been very successful so far. In this example, I’ve created a tolerance of 0.05 meters +- around my target point, and I’ve added it to a model that’s already had a lot of training (I found this overall approach to be disastrous on a model from scratch). Any advice?
if closest_waypoint[1] == 80 and -0.3287 > x > -0.4287 and -1.4798 > y > -1.5798:
reward = 1
elif closest_waypoint[1] == 81 and -0.2747 > x > -0.3747 and -1.5144 > y > -1.6144:
reward = 1
elif closest_waypoint[1] == 81 and -0.2322 > x > -0.3322 and -1.537 > y > -1.637:
…
else:
reward = 0.001