Development of an Autonomous Off-Road Ground Vehicle Simulator

The Objective of this STTR project is to develop the Autonomous Vehicle Simulation Platform (AVSP) as an open-source tool. The AVSP will virtually test how teams of autonomous interconnected wheeled, tracked and legged ground vehicles operating on various kinds of deformable terrains interact with each other and with their environment. The AVSP will simulate the output of various sensors carried by the vehicles including visible light and infrared cameras, Lidar, radar, ultrasonic, GPS, IMU, magnetometer and other sensors that measure the state of the vehicle’s drivetrain and suspension. The AVSP will be able to connect to any vehicle autonomy AI using ROS (Robot Operating System). Inter-vehicle communication between the virtual autonomous vehicles will be implemented using JAUS and DSRC. Thus, vehicle AI designers will be able to test their control algorithms, sensor setup, and inter-vehicle communication strategy in a realistic multi-layered photorealistic Virtual Environment (VE). A GIS importer will be developed that will use GIS data to automatically generate the terrain topography, set the soil and vegetation characteristics, and import geographically referenced CAD models of man-made and natural static objects. The VE will include the visual and physical characteristics of the terrain’s topography, subsurface layers, vegetation, road networks, structures, obstacles, and other agents such as animals, humans and vehicles that are not part of the autonomous team. Designers will also be able to set various environmental conditions in the VE that can vary over time such as rain, fog and snow. Furthermore, designers will be able to adjust the appearance and soil characteristics of the VE depending on the season. The time of year and time of day will also affect the positions and motions of the sun and moon in the sky during the simulation. A scenario controller will allow users to control aspects of the scenario in real-time, setup time or condition based events, view the scenario through a world view or using any of the vehicles’ sensors, and drive one or more of the scenario’s vehicles manually to test the interaction between manned/remote and unmanned autonomous vehicles. Parallel computing will be used to allow the scenarios to run in real-time. A realistic and operationally relevant scenario that was developed by the NATO AVT-341 research task group (RTG) on “Mobility Assessment Methods and Tools for Autonomous Military Ground Systems” will be used to test the AVSP and assess its performance. Finally, a community website will be established to facilitate the free dissemination and further development of the AVSP, and facilitate the interaction between the AVSP users and developers.