search

AIcrowd / real_robots

Gym environments for Robots that learn to interact with the environment autonomously
https://www.aicrowd.com/challenges/neurips-2019-robot-open-ended-autonomous-learning
MIT License
34 stars 18 forks source link

Goal miss match in extrinsic phase #23

Closed tkyaaida closed 4 years ago

tkyaaida commented 4 years ago

I noticed that the first obs['goal'] the controller receives in each extrinsic trial is actually one before. I think this is due to setting a goal after creating observation by resetting environment.

I think simple fix would be something like below: 

    def run_extrinsic_trial(self, trial_number):
        observation = self.env.reset()
        reward = 0
        done = False
        self.env.set_goal()

        observation['goal'] = self.env.goal.retina

https://github.com/AIcrowd/real_robots/blob/3f6cb2db467d0a7c251de1edc2828a3bb82f81dc/real_robots/evaluate.py#L240-L244

spMohanty commented 4 years ago

@tkyaaida : Thanks for pointing it out. Can you please send across a pull request with the described change ?