Exercise 2.1c

Moderator: Lernende Roboter

roberto.calandra
Neuling
Neuling
Beiträge: 5
Registriert: 3. Dez 2013 15:58

Exercise 2.1c

Beitrag von roberto.calandra » 16. Dez 2013 14:42

Dear Students,

Please eliminate the first 949 states of the trajectory to follow, and try to follow only the remaining part.

load('inv_dyn_ref_traj.mat');

states(1:949,:) = [];

The problem is that as somebody pointed out (and i didn't understand at the moment, I apologize), the trajectory is very far away from the training set. Therefore using normal methods, you get very poor predictions.

The reason why this didn't happen with my code during the Exercise lecture next week, is probably that I use more advanced algorithms to avoid numerical problems.

Sorry again for the problem.
cheers
-R

Zurück zu „Lernende Roboter“