Kjell

@kaihuchen Have you looked further in this direction now? I get very great results with the simple score technique and achieve my goals actually. But to extend my mission goals,...

5. Continuous action spaces become necessary. Issue with baselines and multidimensional action spaces is know. Other integration frameworks like keras-rl support multidimensional action spaces.

Workaround to **3.** : I adapted for Multirotor and I am able to run episodes though my `reset `workaround:  ``` def reset(self): reset = False z = self.z while...

Comment to **5**: I do not see this problem. You have to make sure to pass a constant `z` in APIs like `moveByVelocityZ `and `moveByAngle`. Do not make the mistake...

It is definitively trouble with baselines. I am successfully running other gym integrations like [keras-rl](https://github.com/matthiasplappert/keras-rl).

@kaihuchen That seems like a lead, but I have the newest version, `baselines (0.1.4)`. Their github: `version="0.1.4` Which system are you using?

So we have the same set up and I am testing also with the pre-built Windows binaries. Strange. I will try to look further into it. For now I will...

Yes I see. I will just have to wrap my head around that. The unclear part is maybe, that I do not have discrete states and observations like in a...

Sorry for the late answer, I had strange notification settings. Also extending our agents to continuous action spaces is straightforward by applying the same changes which the[ continuous-action version of...