No working example of GPMPC + Learning

[cblancovolle]

It seems that, in current version, there is no working example of GPMPC with learning for quadrotor tasks.
I attempted to run example/mpc with gp_mpc as the algo and adding experiment.launch_training() in mpc_experiment.py.

Here is the full modified mpc_experiment.py file:
mpc_experiment.py

Here is the config file for quadrotor_2D that I used:
gp_mpc_quadrotor_2D_tracking.yaml

When running the experiment (on Windows) I get the following exception:

Traceback (most recent call last):
  File "...\testing\paper-reproduction\safe-control-gym\examples\mpc\mpc_experiment.py", line 121, in <module>
    run()
  File "...\testing\paper-reproduction\safe-control-gym\examples\mpc\mpc_experiment.py", line 41, in run
    experiment.launch_training()
  File "...\testing\paper-reproduction\safe-control-gym\safe_control_gym\experiments\base_experiment.py", line 229, in launch_training
    self.ctrl.learn(env=self.train_env, **kwargs)
  File "...\testing\paper-reproduction\safe-control-gym\safe_control_gym\controllers\mpc\gp_mpc.py", line 876, in learn
    run_results = self.prior_ctrl.run(env=train_envs[0],
AttributeError: 'LinearMPC' object has no attribute 'run'

[Federico-PizarroBejarano]

@adamhall I am unsure how GP-MPC works. How difficult would it be to get an example script working? What is wrong in the example @cblancovolle is using?