Camera configuration on evaluation server

Hello @sidharth_kathpal6,
thanks for the investigation!
I copied the original config.py file to be sure and modified the according part. However, the environment still seems to only give me one image. Is it possible that the change does not work yet?

View post 2

Need your Inputs for improving competition

The cameras on the evaluation server are still not configured according to the rules in the challenge overview, as I described in this thread:

This is currently keeping me from submitting my agent. Are there any news on this topic?

View post 2

Need your Inputs for improving competition

Also it would be super cool if we were allowed to cut the curbs on the track, that would just look absolutely sick

View post 2

Need your Inputs for improving competition

@siddha_ganju @jyotish
Hi, I found a bug in env.py, lines 689-691:

            ) + np.arctan(
                dx / dy
            ) # yaw, radians

This should be:

            ) + np.arctan2(
                dx, dy
            ) # yaw, radians

This is a bug because with arctan the car is facing the wrong way when resetted to some segments, leading to episode termination after the car starts driving. Reference: https://numpy.org/doc/stable/reference/generated/numpy.arctan2.html

View post 1

Camera configuration on evaluation server

Could you please point out where in the rules it says that we only have one camera view during evaluation? Maybe it is the better option to change the server configuration instead of the rules. Thank you very much!

View post 2

Camera configuration on evaluation server

Hi @sidharth_kathpal6,
thank you for the clarification. I think the challenge overview is a bit misleading in that matter. In section “THE learn-to-Race Framework” it says:

During evaluation, the agents will ONLY have access to speed and RGB images from cameras placed on the front, right, and left of the vehicle.

Maybe that should be clarified as well.

Edit:

This is also stated in section “Competition Structure” / “RULES”.

It is just a bit unfortunate because our previous development had these rules in mind.

View post 2

Camera configuration on evaluation server

As I understood, we are allowed to use front-, left- and right-facing RGB cameras during the evaluation of our submissions on the remote server.

I was able to configure those observations in my local setup, but according to the logs, the evaluation environment on the remote server provides me only one camera image in obs[1], i.e. len(obs[1])==1.

How do I enable all three image observations?

View post 2

Need Help with the Simulator documentation

Hello @AnimeshSinha1309,
I think I found a bug that is causing the problem with all images being the same.
In l2r-starter-kit/l2r/envs/utils.py, I changed lines 190-191 from

self.sock.connect(f"{ip}:{port}")
self.addr = f"{ip}:{port}" if not addr else addr

to

self.addr = f"{ip}:{port}" if not addr else addr
self.sock.connect(self.addr)

This resolves the issue since otherwise the socket connects to the default ip and port instead of the passed addr argument.

I hope this helps!

View post 2