This is an autocross simulation engine for MIT AFSAE. For most users, you should run konics using
Docker to ensure consistent behavior. On the other hand, if you need every bit of performance, this
should work out-of-the box on macOS and Windows 10; if you use some other OS, you will need to
compile POV-Ray yourself and make sure the povray binary can be found on your PATH.
Note: The new proton rendering engine is much, much faster than povray. However, it doesn't run
on headless servers yet and is highly experimental. If you want to try it out, run the proton app
in the experimental directory. See this YouTube video
if you run into problems setting it up. Note that if you are doing deep learning on the GPU, you may
want to run this with --blacklist-accelerated-compositing --blacklist-webgl so it doesn't use your
discrete GPU.
Let's break down the below commands. First we pull the latest build of konics from the Docker Hub -
if it says not found, that's probably because it's private and you should email me for access. Next
we run the container with the -p flag which passes port 8888 in the image (jupyter) to port 8888
on our local machine. We also add a shared volume using the -v flag so that our current directory
is available inside the machine as /host.
docker pull kevz/konics:v2
docker run -it -p 8888:8888 -v $(pwd):/host kevz/konics:v2
If you want to get back to basics, you can always run docker run -it kevz/konics:v2 bash and do
linux-y things, just for fun. The konics library is on the PYTHONPATH so you should be able to
import it from any script.
Now that we have the latest and greatest version of konics, let's try importing the Drive object
which provides a car-like interface for moving through the simulated world and the make_wt helper
function which makes it easy to dynamically generate a world and track from a map id.
- Note: Another option is to import the World and Track objects and manually construct a world by dynamically importing the map that you want. However, since dynamic imports are tricky and annoying to use, we provide this helper method for loading the 12 built-in maps. *
from konics.core import Drive
from konics.utils import make_wt
We make a world-track tuple from map "alfa" - see letters A-L in the NATO phonetic alphabet for the other map ids - and then create a Drive object for moving around in our new world. We also get the pose/position at time 0 for the track and move our driver there.
- Note: The time parameter as used in the parametric equations is independent of the actual time it takes to drive around a track. One is a mathematical convenience, the other is how you win. *
world, track = make_wt("alfa")
drive = Drive(world)
drive.set_pose(track.get_pose(0.0))
Now that our world is ready and our driver is in place, we can start driving through the track. The below code alternates between turning slightly left and right, driving forwards and taking pictures after each movement.
from scipy.misc import imsave
drive.move(0.1)
for i in range(10):
imsave(str(i) + ".png", drive.render())
drive.move(-0.2 if i % 2 == 0 else 0.2)
The end.