Watch ON Demand
This session is only available to AutoSensLEARN Conference ticket holders.
If you wish to upgrade from your AutoSens to an AutoSensLEARN ticket please email email@example.com.
Login to view this session.
Don't have a ticket yet?
Buy a full AutoSensLEARN conference pass and access ALL content from Detroit and Brussels editions of AutoSens (over 150 sessions!), plus 12 hours+ of in-depth technical tutorials for just £639
It is not possible to evaluate the performance of an autonomous system using only on-the-road trials. Simulation provides an alternative method for exposing autonomous systems to the millions of potential edge cases that will arise in wide-spread autonomous operations. There are numerous commercial simulation products available for development and evaluation of on-the-road autonomous systems. There are fewer solutions available for simulating off-road, unstructured environments. The workshop will review basic information about simulators for autonomous vehicle development and testing, delve into the challenges associated with simulation, particularly related to neural networks and artificial environments and in off-road environments for both commercial and military ground vehicles. The workshop will include a hands-on opportunity with the Mississippi State University Autonomous Vehicle Simulator, an open-source, physics-based simulation library for autonomous vehicles.
A pre-compiled version of the Mississippi State University Autonomous Vehicle Simulator (MAVS) is available for tutorial attendees at: https://www.dropbox.com/s/4eifnc8tdbzvjya/mavs_windows10_06Nov2020.zip?dl=0
This version of MAVS is compiled for Windows 10 and requires Python with numpy libraries to be installed.
The ZIP archive includes the pre-compiled MAVS software. I usually install it in C:\dev\ and will be working in a folder in a similar location during the tutorial.
The ZIP archive includes a README.txt with setup instructions as well as a set_up_mavs_instructions.pptx that provides more details and screenshots on the setup process.
There is a simple simulation_example that attendees can test as part of the setup to ensure MAVS is working.
Finally, there is a MavsPythonUserGuide.pdf that provides more information on using MAVS and the Python API.
While the tutorial will use MAVS, MAVS is not the focus of the tutorial. We will discuss broader issues in modeling and simulation of off-road autonomous ground vehicles, current state of the art in tools (including but not limited to MAVS), and current work developing tools. We will use MAVS to explore some of the steps required to simulate AGVs and to demonstrate some applications and data collection.