OC vs. Neurocontroller (+noise)
The trajectories as given by the optimal control solution are run in the simulation here (with and without Noise in the integrator on \ddot x and \ddot y) and are compared with the neurocontrollers that performed well.
| Max. Noise | OC | NC (w\o noise) | NC (noise 0.2, one run) | NC (noise 0.2, 5 runs) |
|---|---|---|---|---|
| ± 0.1 (10 runs) |
 |
 |
 |
|
| ± 0.2 (10 runs) |
f: -0.5528 |
f: -0.4814 |
f: -0.4812 |
f: -0.7018 |
| ± 0.3 (10 runs) |
 |
 |
 |
|
| ± 0.4 (10 runs) |
 |
 |
 |
Optimal Control Trajectories
The trajectories as given by the optimal control solution are run in the simulation here (with and without Noise in the integrator on \ddot x and \ddot y).
| Max. Noise | -2 / 0 | -2 / 2 |
|---|---|---|
| ± 0.0 |  |
 |
| ± 0.1 (10 runs) |
|
 |
| ± 0.2 (10 runs) |
|
 |
| ± 0.3 (10 runs) |
|
 |
| ± 0.4 (10 runs) |
|
 |
| ± 0.5 (10 runs) |
 |
 |
| ± 1.0 (10 runs) |
 |
 |
Selected Trajectories
Here are some interesting plots that came out during the evolution of the neurocontroller.
