In this project a car is made to navigate autonomously in a simulated environment. 9 layer convolution neural network is implemented to clone autonomous driving behavior
Code for use in my autonomous driving research. Consists of largely-independent python packages making heavy use of the numpy, scipy, and matplotlib libaries.
-
hgmm:
A package implementing Frank Havlak's hybrid gaussian mixture model algorithm.
Havlak, Frank, and Mark Campbell. "Discrete and continuous, probabilistic anticipation for autonomous robots in urban environments." IEEE Transactions on Robotics 30.2 (2014): 461-474.
-
collision:
A package implementing Jason Hardy's fast collision probabilty algorithm between two convex polygonal objects, one of which has an uncertain position.
Hardy, Jason, and Mark Campbell. "Contingency planning over probabilistic obstacle predictions for autonomous road vehicles." IEEE Transactions on Robotics 29.4 (2013): 913-929.
-
myplot:
Helper package for plotting gaussians as heatmaps or covariance ellipses