Inspiration

Accident scenes disappear within minutes due to traffic clearance, weather, and crowd movement. Traditional investigations rely on manual sketches and estimation, often leading to errors. We wanted a faster, more accurate, and technology-driven way to preserve accident evidence before it is lost.

What it does

AARS is a mobile-first AR system that scans accident scenes using a smartphone and reconstructs them in accurate 3D. It detects vehicles, skid marks, and road geometry, runs physics-based collision analysis, and generates court-ready digital reports with AR replay.

How we built it

We designed a guided AR scanning flow using ARCore/ARKit concepts. Captured frames are processed with computer vision and AI for object detection, followed by 3D reconstruction using photogrammetry or LiDAR. Physics models estimate collision dynamics, and the results are visualized through AR and reports.

Challenges we ran into

Handling varying lighting conditions, ensuring scale accuracy, and balancing realism with performance on mobile devices were major challenges. Designing a simple flow usable by officers under real-world constraints was also difficult.

Accomplishments that we're proud of

We built a complete end-to-end workflow—from scan to reconstruction, simulation, AR replay, and report generation—focused entirely on real-world usability and impact.

What we learned

We gained hands-on experience with AR workflows, computer vision pipelines, 3D reconstruction concepts, and physics-based simulation, while learning how to design for field-level constraints.

What's next for AARS – Augmented-Reality Accident Reconstruction System

We plan to improve model accuracy, add real-time collaboration, integrate with police and insurance systems, and deploy AARS as a scalable, production-ready forensic tool.

Built With

  • ar-foundation
  • arcore
  • arkit
  • arkit-computer-vision-&-ai:-yolo
  • aws-s3-/-firebase-storage
  • aws-s3-/-firebase-storage-visualization-&-reporting:-unity3d
  • camera-&-sensor-apis
  • fastapi-/-node.js
  • flutter-/-react-native
  • javascript
  • javascript-mobile-/-ar:-flutter-/-react-native
  • lidar-apis
  • lidar-depth-apis
  • open3d
  • open3d-physics-simulation:-vehicle-dynamics-models
  • opencv
  • pdfkit
  • photogrammetry-(colmap)
  • physics-simulation-models
  • postgresql
  • python
  • reportlab
  • reportlab-/-pdfkit-security-&-deployment:-encryption
  • rest-apis
  • scipy-/-bullet-physics-backend-&-apis:-fastapi-/-node.js-database-&-storage:-postgresql
  • tensorflow-lite
  • tensorflow-lite-3d-reconstruction:-photogrammetry-(colmap)
  • unity
  • yolo
Share this project:

Updates

Submission history