How Co-Simulation Accelerates the Future of Autonomous Driving

Dr. Thomas Benz, Solution Director Automotive, PTV Group

For autonomous vehicles to be able to handle any scenario that could happen on a public road, extensive testing and validation is required. It is not only time-consuming, but also cost-intensive endeavor to reach the needed number of test miles and thus ensure safety without driver interventions. Such autonomous vehicles have to manage traffic, pedestrians and cyclists, and changing weather conditions. The control algorithms are fed with huge amounts of data. If all test driving would be done on real roads, this process could easily take several years to be completed. An impossible task for manufacturers. To take the fast lane to autonomous safety and reliability, simulation is the key: Vehicle manufacturers can run thousands of different scenarios in virtual test beds that equal hundreds of miles driven on real roads to reach validation and go-to-market faster.

From simulation to co-simulation

Simulation is an efficient way to supplement real-world testing, but also comes with some challenges. Replicating, for example, specific road conditions, driver behaviors, traffic schemes and motion characteristics is complex and every aspect requires specific expert knowledge that needs to be connected in the right way. In order to design such an environment that meets all the requirements of an ideal test bed for autonomous vehicles, you need to consider various driving and driver dynamics, covering sensor and road network information, different scenarios for 3D models and traffic signals, while also including the surrounding traffic and V2X communication.

This is why a tool chain of different simulation systems is the solution. Each one of these systems implements specific know-how and together they form a pool of knowledge. Simulation tools are connected to form co-simulation: The tools run synchronously and exchange data during run time so that one system reacts to the behavior of another. The joint solution simulates real test scenarios, which do not only represent the behavior of a single vehicle, but also of the vehicle’s environment and the immediate reaction of other road users. To be able to make the necessary adjustments to the vehicles, development teams profit from the information gained through co-simulation of dedicated simulation products.

Virtual testing based on real traffic models

Simulation enables manufacturers to reproduce traffic conditions and test innovations without being affected by external conditions. PTV Vissim offers a solution for modeling individual vehicles and traffic flows in a highly realistic manner. Here the effort is the same if you want to simulate a car with a background traffic of 100 or 1000 vehicles. Standard applications and existing transport networks are instantly available as well.

Manufacturers can run their own vehicle models in the co-simulation, for example with PTV Vissim. The tool allows users to model scenarios for specific times of the day, but also for different traffic and weather conditions. For example – based on real traffic models – manufacturers can easily create a simulation environment that reflects traffic conditions at a certain time of day, in a specific city on a specific day – from a relatively calm Sunday morning at 8:00 am in Washington D.C. to a busy evening peak at 6:00 pm on a Monday night in Taichung. Any traffic situation, be it on a motorway or on an inner-city road can be simulated, pathing the way for a safer and more reliable future of autonomous vehicles.