eazydammy/unscented-kalman-filter-vehicle-tracking

State estimation and tracking of multiple vehicles on simulated highway traffic by fusing noisy lidar and radar measurements using an Unscented Kalman Filter.

Unscented Kalman Filter Vehicle Tracking

In this project, an Unscented Kalman Filter is implemented to track and estimate the state of multiple cars on a highway using noisy lidar and radar measurements as shown.

• main.cpp is using highway.h to create a straight 3 lane highway environment with 3 traffic cars and the main ego car at the center.
• The viewer scene is centered around the ego car and the coordinate system is relative to the ego car as well. The ego car is green while the other traffic cars are blue.
• The traffic cars will be accelerating and altering their steering to change lanes.
• Each of the traffic car's has it's own UKF object generated for it, and will update each indidual one during every time step.
• The red spheres above cars represent the (x,y) lidar detection and the purple lines show the radar measurements with the velocity magnitude along the detected angle.
• Tracking is along the X/Z axis only. The Z axis is not taken into account for tracking.

---

Other Important Dependencies

• cmake >= 3.5
  • All OSes: click here for installation instructions
• make >= 4.1 (Linux, Mac), 3.81 (Windows)
  • Linux: make is installed by default on most Linux distros
  • Mac: install Xcode command line tools to get make
  • Windows: Click here for installation instructions
• gcc/g++ >= 5.4
  • Linux: gcc / g++ is installed by default on most Linux distros
  • Mac: same deal as make - install Xcode command line tools
  • Windows: recommend using MinGW
• PCL 1.2

Basic Build Instructions

1. Clone this repo.
2. Make a build directory: mkdir build && cd build
3. Compile: cmake .. && make
4. Run it: ./ukf_highway

Generating Additional Data

Custom radar and lidar can be generated by modifying the code in highway.h to alter the cars. Also, tools.cpp can be modified to change how measurements are taken, for instance lidar markers could be the (x,y) center of bounding boxes by scanning the PCD environment and performing clustering.