Enhancing Autonomous Vehicle Performance Through Simulation
A lot of test simulations are needed to ensure the safety of autonomous vehicle systems. SIMCenter researchers are making it faster and easier with an open-source software tool.
Challenge
A team of SIMCenter researchers seeks to develop an open-source software tool that can be used by the public to create any number of test scenarios to validate the expected behavior of advanced driver-assistance systems (ADAS), and then share those validation scenarios with others.
Solution
Combining the traffic scenario generation of SUMo (Simulation of Urban Mobility) with the 3-D rendering capabilities of Unreal Engine, the open-source software tool can simulate both the world around a vehicle, the vehicle itself, and the vehicle's controllers. Moreover, the software tool features vehicle dynamics, traffic simulations, sensor models, and plug-and-play connections for ADAS controllers.
Results
The software tool grants users the freedom to develop their own unique validation scenarios from scratch, and its open-source nature allows for financial savings, community participation, and enhancement of the tool's components through collective competition.
Summary
At the forefront of autonomous vehicle development are researchers like SIMCenter Director Dr. Shawn Midlam-Mohler, who is joined by Dr. Punit Tulpule, Aditya Karumanchi, and Yishen Jin in an ongoing project to improve autonomous vehicle performance through simulations. Partnering with Ford and Honda, the SIMCenter team has created an open-source software tool that is substantially more cost effective than alternatives, resulting in savings between $20,000-$40,000 per software license. Based on software like SUMo (Simulation of Urban Mobility) and Unreal Engine, the tool is capable of generating traffic scenarios and rendering them in 3-D. Those features, along with the tool's open-source nature, means that it can be used by anyone to test and analyze the expected behavior of advanced driver assistance systems (ADAS) in autonomous vehicles.
In addition to the financial savings, the software tool also comes with a number of benefits. For instance, the simulations can be completed in less time than on-road tests, thereby accelerating the testing process. Another key benefit is that the users have the freedom to develop their own scenarios from scratch, all of which will be added to a growing pool of test scenarios. As a result, the tool is continuously enhanced by community participation and collective competition. The sum total of these benefits is a cost effective, community-driven instrument for the ongoing advance of autonomous vehicle research.