Physics-Based Sensor Models for Virtual Simulation of Connected and Autonomous Vehicles

A C++ library for multi-physics simulation developed as part of the SAFER-SIM project. The Chrono::Sensor module augments the Project Chrono engine to generate synthetic sensor data for cameras, lidar, GPS, and IMU using ray-tracing algorithms. The project report includes results from sensor implementations and demonstrations of autonomous vehicle simulation.

Use Cases

• Simulating camera data for perception system testing based on the ray-tracing camera model
• Generating synthetic lidar point clouds for obstacle detection algorithms based on the described lidar model
• Testing sensor fusion algorithms using the co-simulation of GPS and IMU data mentioned in the framework
• Developing custom sensor models by leveraging the provided framework for implementing new sensors

Strengths

• Framework supports modeling of multiple sensor types: camera, lidar, GPS, and IMU
• Sensor data generation uses physics-based ray-tracing algorithms mimicking real acquisition
• Includes functionality for adding sensor-specific noise for data augmentation
• Open-source and available as a module for the established Project Chrono multi-physics engine

Limitations

• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment for specific ML tasks
• Last update date is unknown; freshness unverified

Provenance

Source

University of Wisconsin–Madison, Project Chrono repository

Collection Method

Developed as a software module for the Project Chrono multi-physics engine

Time Range

null

Freshness

null

Geography

null