Edouard Leurent

Hi, Try as indicated in the docs: ``` import gym import highway_env env = gym.make("intersection-v0") env.configure({ "manual_control": True, "real_time_rendering": True, }) env.reset() done = False while not done: _, _,...

Hi, ~~No this is not a known issue, and I could not reproduce it with your code snippet. When I run it, the RAM stays fairly constant through time. Do...

you don't need to import anything, if you have access to a vehicle you can directly check its target lane index. For instance: `vehicle = env.vehicle` or `vehicle = env.road.vehicles[0]`...

Then `print(env.vehicle.target_lane_index)`

yes

Interesting, apparently gym==0.17.3 is installed by default on Colab, but the RecordVideo wrapper only appeared on gym==0.21 or 0.22. I recently made it a required dependency of highway-env, but I...

I have only tried to use openai/baselines, DLR-RM/stable-baselines3, and eleurent/rl-agent libraries. But highway-env follows the common openai/gym interface, so if these DRL libraries support it then it should work fine....

- For simple kinetic constraints (position, speed, acceleration, curvature), you should edit vehicle/kinematics.py - For dynamical constraints (jerk; variance curvature; etc) you should edit vehicle/dynamics.py - And for the min...

Yes, that is what happens when you set the action type to [DiscreteMetaActions](https://highway-env.readthedocs.io/en/latest/actions/index.html#discrete-meta-actions) (which is the default). The RL agent's actions set the speed and lateral position setpoints, which are...

You can select several options: By using ``` env = gym.make('highway-v0') env.configure({ "action": { "type": "ContinuousAction" } }) env.reset() ``` you get continuous actions where you can set a and...