https://github.com/sunsided/ballistics

A Kalman filter ballistic curve estimation playground
https://github.com/sunsided/ballistics

ballistic-curve kalman-filter monte-carlo-sampling rust simulation

Last synced: 11 days ago
JSON representation

A Kalman filter ballistic curve estimation playground

• Host: GitHub
• URL: https://github.com/sunsided/ballistics
• Owner: sunsided
• Created: 2024-06-08T17:27:15.000Z (about 2 years ago)
• Default Branch: main
• Last Pushed: 2026-05-04T20:35:05.000Z (about 2 months ago)
• Last Synced: 2026-05-04T22:26:14.652Z (about 2 months ago)
• Topics: ballistic-curve, kalman-filter, monte-carlo-sampling, rust, simulation
• Language: Rust
• Homepage:
• Size: 211 KB
• Stars: 0
• Watchers: 1
• Forks: 1
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • Agents: AGENTS.md

Awesome Lists containing this project

README

          
# Ballistics Estimation

[![CI](https://github.com/sunsided/ballistics/actions/workflows/ci.yml/badge.svg)](https://github.com/sunsided/ballistics/actions/workflows/ci.yml)

[![license: EUPL-1.2](https://img.shields.io/badge/license-EUPL--1.2-blue.svg)](https://github.com/sunsided/ballistics)

A small Rust/ggez playground that tracks a wind-perturbed projectile with a Kalman filter and predicts its impact zone on the ground via Monte-Carlo sampling of the filter's covariance. Acts as a testing ground for the [minikalman](https://github.com/sunsided/minikalman-rs) crate.

![Screenshot](docs/screenshot.png)

*Blue = true firing trajectory (gravity only), Green = filter-predicted trajectory, Cyan dot + ellipse = state estimate with 1σ covariance, Orange bar + whiskers = predicted impact zone (±2σ with min/max).*

## What it does

- An opponent on the right edge fires projectiles with a random launch angle drawn from Normal(60°, 10°) clamped to 25°–80° and a speed of 100–120 m/s.

- Gravity (-9.81 m/s²) and a time-correlated wind model (sinusoidal base plus Ornstein–Uhlenbeck-style correlated noise on the x-axis) perturb the flight.

- The observer receives noisy 2D position measurements each physics tick (uniform noise in ±5 px per axis).

- A 6-state linear Kalman filter with state vector `[x, y, vx, vy, ax, ay]` (constant-acceleration model, dt-based state transition) is built on `minikalman`; the measurement matrix observes position only.

- Impact prediction runs 64 Monte-Carlo rollouts per update on a background thread: it Cholesky-decomposes the 4×4 position/velocity sub-covariance, draws Gaussian samples, propagates each sample until it crosses the ground (with a quadratic root-solve for the exact crossing time), and aggregates mean, std-dev, min, and max of the landing x to draw the impact zone.

- Game time is accelerated by `GAME_TIME_FACTOR = 10.0` at `PHYSICS_FPS = 60` for faster iteration.

## Running it

```bash

cargo run --release

```

The `resources/` directory containing `LiberationMono-Regular.ttf` must be present, as `main.rs` adds it as a resource path for text rendering. The font is stored via [Git LFS](https://git-lfs.com/); make sure Git LFS is installed (`git lfs install`) before cloning, or run `git lfs pull` after cloning.

## Ideas

- Can we estimate the mass of the projectile as well?

  - Both the mass and the initial firing force determine the acceleration.

  - The mass will affect the curvature of the trajectory.

## License

Licensed under the [European Union Public Licence v1.2 (EUPL-1.2)](https://joinup.ec.europa.eu/collection/eupl/eupl-text-eupl-12).