ISimA: Intelligent Simulation for Testing and Training of Autonomous Single- and Multi-Vehicle Systems
Abstract
Major Goals: The ISimA project was designed to develop a state-of-the-art software simulation system leveraging commercial off-the-shelf simulation engines to allow for the testing, training, and evaluation of autonomous single- and multi-vehicle systems. In Phase I of the project, we focused on the following three questions in order to prove the feasibility of the overall architecture: 1. The first question is whether it is feasible to build a single simulation tool that will provide adequate fidelity of simulation for testing and training autonomous single- and multi-vehicle systems operating off-road. In particular, the simulator will need to provide proper simulation of sensing (both LIDAR and RGB camera), vehicle dynamics, vehicle-to-terrain interaction, limited communication bandwidth between multiple vehicles, and simulation of the behavior of the virtual world. 2. The second question is whether it is feasible to develop an intelligence engine for generating test scenarios and dynamic control of dynamic entities within ISimA to explicitly optimize the level of confidence in the ability of autonomous vehicles to achieve their designated tasks. 3. The third and the final question is whether ISimA can be built on top of open-source software under proper licenses that will allow for unrestricted use of and the ability to extend ISimA by others.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 30, 2021
- Accession Number
- AD1187975
Entities
People
- Jonathan Butzke