- Virtual simulation accelerates development for advanced driver assistance systems (ADAS) and automated driving (AD)
- Virtual tests help to reduce risks and offers full test recording capabilities for precise identification for software improvement
- Testing up to nearly 5,000 miles per hour is possible with virtual simulation
Las Vegas, January 9, 2019 - No compromise on safety, this is what the technology company Continental stands for. Through a unique collaboration with the German based start-up company Automotive Artificial Intelligence (AAI), Continental enhances its approach for virtual development of its ADAS and AD functions. Virtual development is already an integral part in today’s product development.
For the development of AD ready functions, a more complex - even real-world - virtual development and test environment is needed. In particular, this applies to the complex simulation of dynamic road-user behavior. The reason simulation with different driving styles (e.g. aggressive, defensive) is important for approximating reality more closely and avoiding unrealistically monotonous, predictable processes. Continental is responding by developing a highly scalable, highly modular development and testing environment that meets the new requirements. AAI has the competence to create virtual environments based on HD-maps and AI-driven road-users, with the intention of doing justice to the increasingly broader scope of validation in automated driving. The simulation tools developed within the collaboration will be made available to the entire automotive industry.
Safer, faster and more cost-efficient
“Our ultimate goal is always absolute safety. This is especially true for automated driving,” said Amit Kapoor, head of the business unit Advanced Driver Assistance Systems at Continental, North America. “Resources in vehicle test are limited and virtual testing can provide fast feedback for developers. While a real vehicle can drive around 6,500 test miles in a physical test in one month, today up to nearly 5,000 miles per hour are possible with virtual simulation. This number will increase in future.”
Set at a level that guarantees the required level of safety, the number of test miless can be achieved faster and more cost-effectively. Real testing continues to be essential for driving function testing, since testing on a purely virtual level is not enough. Virtual tests help to reduce risks before real world testing and offers full test recording with data analytics and incident replay capabilities for precise identification of software improvement.
AAI and Continental have already begun their collaboration. “This positions Continental as a pioneer in the field of virtual simulation tools for the automotive market,” Kapoor said.
Two companies, two environments, one common aim
The collaboration sets up a clear separation between vehicle simulation and environment simulation. AAI is responsible for the environment, providing software that enables driving vehicles in a virtual world. Individual parameters determine the weather, traffic infrastructure, road markings, and objects such as road signs and potholes. As a second element, AAI uses artificial intelligence to integrate traffic participants (agents) into the virtual simulation environment and uses machine learning algorithms to train agent behavior, resulting in aggressive, moderate and defensive driver profiles. The long-term goal is to create a “replica of the world” and realistically simulate all road users and environmental factors.
Continental is responsible for vehicle simulation and is building a virtual vehicle with software that masters various driving functions at different levels of automation. Continental's task still includes completely representing its own vehicle and all the software components it contains, with the simulation task of driving dynamics. The entire ADAS product portfolio will be tested.
Since AAI and Continental are openly collaborating with each other, the interfaces of both environments are interconnected. “The collaboration makes our development cycles shorter and safer,” Kapoor said. “A big advantage of virtual simulation is test reproducibility and that a lot more use-cases can be tested virtually than we can in real-world testing.”
Collaboration with impact: Faster test procedures will bring automated driving on the road
All partners will benefit equally from the collaboration. Continental will gain software that can create realistic traffic simulations, enabling new products in the ADAS portfolio to be tested virtually. AAI increases its automotive sector and sensor technology expertise.
“For our young company, partnering with a technology corporation like Continental is a seal of quality,” said Isabel Metz, head of Business Development at AAI. “Continental's customers will also benefit because a faster test procedure guarantees safer, more robust and more cost-effective products.”
Continental develops pioneering technologies and services for sustainable and connected mobility of people and their goods. Founded in 1871, the technology company offers safe, efficient, intelligent and affordable solutions for vehicles, machines, traffic and transport. In 2017, Continental generated sales of €44 billion and currently employs more than 244,000 people in 61 countries and markets.
Automotive Artificial Intelligence (AAI) GmbH is a software start-up founded in February 2017 that aims to accelerate the development of autonomous vehicles. The target group are engineers in the automobile industry and system suppliers. Using AAI’s platform, these engineers can create detailed simulations, with programmable traffic situations and road conditions, in order to test their vehicles in endurance tests and other specific scenarios. The simulation runs in realistic environments that have been created based on high definition street maps. Today a virtual vehicle only needs 8 hours to travel the same distance that would take a real-world car more than one month. The aim is to further widen this gap - by increasing the simulation speed and scaling on the one hand and on the other by operating the simulation in a customer cloud. The system is self-learning and works with artificial intelligence. The precise reproducibility of different traffic situations as well as the reduced testing time are the key advantages for developers. Currently the company employs more than 50 people of more than 15 nations.