radar-filter-cpp

A C++17 alpha-beta tracking filter, written as a command-line tool. It is a C++ version of the alpha-beta filter idea from my JavaScript RadarTracker, with its own gains (alpha 0.5, beta 0.1) and tests.

What it does

Reads noisy 2D position measurements ('t,x,y') from a CSV file, runs one alpha-beta filter on x and one on y, and prints the smoothed position and estimated velocity at each step.

Each update has three parts:

1. Predict where the target should be, from the last position and velocity.
2. Measure the residual: how far the new reading is from the prediction.
3. Correct position (by alpha) and velocity (by beta) toward the reading.

Build and run

g++ -std=c++17 -Wall -Wextra alpha_beta.cpp main.cpp -o tracker
./tracker measurements.csv

Test

g++ -std=c++17 -Wall -Wextra alpha_beta.cpp test_alpha_beta.cpp -o test_alpha_beta
./test_alpha_beta

Tests check a stationary target, learning a constant velocity, damping a noise spike, and rejecting invalid settings. They run on every push with GitHub Actions.

Design notes

• Filter logic lives in its own class (alpha_beta.hpp/.cpp), separate from file input/output, so it can be tested on its own.
• Invalid settings (alpha outside (0,1], beta outside (0,2], dt not positive) throw std::invalid_argument.
• No external libraries; standard C++17 only.

What I'd add next

A full Kalman filter (adaptive gains from noise covariances), CMake, and GoogleTest.