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

This document contains the final project report for the SAFER-SIM project titled “Physics-Based Sensor Models for Virtual Simulation of Connected and Autonomous Vehicles.” The report includes discussion of sensors models for simulation autonomous vehicles, and overviews the simulation framework developed in accordance with the project. The framework, called Chrono::Sensor is developed as a module alongside Project Chrono to augment the open-source multi-physics engine with the capability to simulation sensor data from within its virtual environment. Chrono::Sensor provides support for the modeling and simulation of camera, lidar, GPS, and IMU. It also provides a framework to implement custom sensors that can leverage existing sensor generation functionality. Chrono::Sensor generates data using ray-tracing algorithms that mimic the data acquisition process of cameras and lidars. In addition to data collection, the sensors support further data augmentation including the addition of sensor-specific noise. Results from each sensor implementation are included as part of the corresponding sensor discussion, with the report concluding with two demonstrations showing the use of Chrono::Sensor, in combination with Chrono, to simulate autonomous vehicles. The dataset for this project is a C++ library for multi-physics simulation. It is available at the following URL: https://github.com/projectchrono/chrono