nplan / gym-line-follower Goto Github PK

does it support cuda and tf-gpu?
and great work by the way

1. pip3 install -e gym_line_follower
   I used 'gym-line-follower' instead of 'gym_line_follower' because the folder name is 'gym-line-follower', and the containing folder 'gym_line_follower' doesn't have setup.py

2. I'm not sure if it matters but I used pip instead of pip3. I only have pip and it corresponds to python3.7

Hi!
Is there any example on how to run the env in render='human' mode?

Thanks!

Hi, this looks like a really cool environment. I was planning to use it in some experiments when I noticed that with the default settings it is rather slow. For example the following code runs at about 67 steps (frames) per second:

env = LineFollowerEnv(gui=False, nb_cam_pts=8, max_track_err=0.4, power_limit=0.4, max_time=600, obsv_type="points_latch")

start, steps = time.time(), 0
env.reset()
for ep in range(20):
     for i in range(1000):
         obsv, rew, done, info = env.step((np.random.rand(), np.random.rand()))
         if done:
             print(ep, i)
             steps += i
             break
    env.reset()
env.close()

print("Did {} steps in {:3.1f}s along {} episodes.".format(steps, time.time() - start, ep))

Did 1931 steps in 28.8s along 19 episodes.

Any suggestion which arguments to change for higher fps without messing the simulation? Thank you!