Self-Driving Car Engineer Nanodegree Program

Project can be compiled with:

$ mkdir -p build && cd build && cmake .. && make

MSE on the provided dataset obj_pose-laser-radar-synthetic-input.txt:

$ ./UnscentedKF ../data/obj_pose-laser-radar-synthetic-input.txt ../data/obj_pose-laser-radar-synthetic-output.txt
Radar enabled, Lidar enabled
RMSE
0.0854383
 0.098572
  0.19357
 0.221618

This satisfies the requirement RMSE <= [.09, .10, .40, .30] in the project rubic.

The algorithm is implemented in ukf.cpp:

• The initialization is done between lines 297-318 of ukf.cpp

  Initialization of the p_x and p_y component is easy. I initialized the velocity to be 5 m/s^2 because it sounds reasonable.

  The covariance matrix P is initialized with small value everywhere, to allow some variations in the states in the first few measurements. The diagonal is increased a bit, depending on the type of the sensor. Laser tends to be more accurate than radar in term of position, so we can use a smaller value to initialize if the first measurement is from the lidar.

• The predict step is done between lines 106-192 of ukf.cpp

  I first generate the augmented sigma points, then predict the sigma points, and update the state mean and covariance.

• The update step is done between lines 198-295 of ukf.cpp

The program can be instructed to use only data from radar, lidar or both, using the third argument:

$ ./UnscentedKF
Usage instructions: ./UnscentedKF path/to/input.txt output.txt [r|l|rl]
 The 3rd parameter specifies the type of sensor to use: r(adar) or l(idar) or rl (both) Default value is rl

The normalized innovation squared (NIS) score is below the 95% threshold most of the time:

Here are the comparision of the location and velocity between the predictions and the groundtruth:

• cmake >= v3.5
• make >= v4.1
• gcc/g++ >= v5.4

1. Clone this repo.
2. Make a build directory: mkdir build && cd build
3. Compile: cmake .. && make
4. Run it: ./UnscentedKF path/to/input.txt path/to/output.txt. You can find some sample inputs in 'data/'.
   • eg. ./UnscentedKF ../data/obj_pose-laser-radar-synthetic-input.txt

We've purposefully kept editor configuration files out of this repo in order to keep it as simple and environment agnostic as possible. However, we recommend using the following settings:

• indent using spaces
• set tab width to 2 spaces (keeps the matrices in source code aligned)

Please stick to Google's C++ style guide as much as possible.

This is optional!

If you'd like to generate your own radar and lidar data, see the utilities repo for Matlab scripts that can generate additional data.

This information is only accessible by people who are already enrolled in Term 2 of CarND. If you are enrolled, see the project page for instructions and the project rubric.