AN OVERVIEW OF A CONNECTED AUTONOMOUS VEHICLE EMULATOR (CAVE)

摘要

We discuss an early version of an open-source autonomous vehicle simulation framework whereby piloting control programs (PCPs) and vehicle response can be evaluated and improved in a safe environment. Through the interaction of hundreds of autonomous and avatar agents in a simulated environment, edge cases for self-driving vehicles can be analyzed thus accelerating research, development and technology deployment. The Connected Autonomous Vehicle Emulator (CAVE) builds upon four foundational components: (i) physics engine and dynamic vehicle support using Chrono and Chrono::Vehicle; (ii) virtual sensors support to provide self-driving algorithms with realistic data; (iii) multi-agent and vehicle-to-vehicle simulated communication support through a fast, low-latency server-client model; and, (iv) virtual environment for both physics and sensing to support edge cases for self-driving vehicles where physical testing is not feasible, including varying environmental conditions such as snow, rain, or fog.