AWS Networking Organisation Internal DeepRacer Cup 2020
Training model | Car | reward function | Gradient descent batch size | Number of epochs | Learning rate | Entropy | Discount factor | Loss type |
---|---|---|---|---|---|---|---|---|
v1-0 | Bennyhwa1 | centre_line | 512 | 10 | 0.0003 | 0.01 | 0.999 | Huber |
v1-1 | Bennyhwa1 | inside_lane_fast | 512 | 10 | 0.0003 | 0.01 | 0.999 | Huber |
v1-2 | Bennyhwa1 | centre_line | 256 | 10 | 0.0001 | 0.01 | 0.999 | MSE |
v1-3 | Bennyhwa1 | inside_lane_fast | 512 | 10 | 0.0001 | 0.01 | 0.999 | MSE |
v1-4 | Bennyhwa1 | inside_lane_safety | 512 | 10 | 0.0001 | 0.01 | 0.999 | Huber |
v1-5 | Bennyhwa1 | fast_straight_line | 512 | 10 | 0.0008 | 0.01 | 0.999 | Huber |
v1-5-1 | Bennyhwa1 | inside_lane_faster | 512 | 10 | 0.0001 | 0.01 | 0.999 | MSE |
v2-0 | Bennyhwa1 | deep_racer_evaluate | 512 | 10 | 0.0003 | 0.01 | 0.999 | Huber |
v2-1 | Bennyhwa1 | deep_racer_evaluate | 512 | 10 | 0.0003 | 0.01 | 0.999 | Huber |
v1-6 | Bennyhwa1 | deep_racer_evaluate | 256 | 10 | 0.0001 | 0.01 | 0.999 | Huber |
v2-2 | Bennyhwa1 | centre_line | 512 | 10 | 0.0003 | 0.01 | 0.999 | Huber |
v1-5-2 | Bennyhwa1 | inside_lane_safety | 512 | 10 | 0.00002 | 0.01 | 0.999 | MSE |
v1-5-3 | Bennyhwa1 | inside_lane_fast | 256 | 10 | 0.00002 | 0.01 | 0.999 | Huber |
v2-3-2 | Bennyhwa1 | inside_lane_fast | 512 | 10 | 0.001 | 0.01 | 0.999 | Huber |
v1-3: 15.8s with 3 off-track.
v1-4: 12.402s with 0 off-track (2nd submission one was 13.364s with 1 off-track, 1st submission was 13.5 with 1 off track, for some reason it just didn't go off track and time improved)
v1-3:
1 00:00:13.056 100% Lap complete
2 00:00:12.508 100% Lap complete
3 00:00:12.360 100% Lap complete
v1-4:
1 00:00:12.474 100% Lap complete
2 00:00:12.784 100% Lap complete
3 00:00:13.225 100% Lap complete
v1-5:
1 00:00:12.971 100% Lap complete
2 00:00:13.434 100% Lap complete
3 00:00:13.324 100% Lap complete
v1-5-1:
1 00:00:01.306 2% Off track
2 00:00:13.252 100% Lap complete
3 00:00:12.946 100% Lap complete
v-1-5-2
1 00:00:01.434 3% Off track
2 00:00:12.929 100% Lap complete
3 00:00:12.822 100% Lap complete
v2-0:
1 00:00:06.533 50% Off track
2 00:00:07.065 54% Off track
3 00:00:12.401 100% Lap complete
v1-6:
1 00:00:12.931 100% Lap complete
2 00:00:13.646 100% Lap complete
3 00:00:13.025 100% Lap complete
v2-1:
1 00:00:01.240 1% Off track
2 00:00:12.328 100% Lap complete
3 00:00:05.561 37% Off track
v2-2
1 00:00:01.527 3% Off track
2 00:00:05.155 35% Off track
3 00:00:05.343 35% Off track
Sensor(s): Camera.
Neural network topology: 3 Layer CNN.
Action space:
- Speed: 1 m/s
- Steering angle: 30°
Sensor(s): Camera.
Neural network topology: 3 Layer CNN.
Action space:
- Speed: 4 m/s
- Steering angle: 30°