Autonomous Vehicles Find a New Playground with Cognata
Mitchell Gracie posted on April 22, 2019 |
Cognata is a lifecycle solution for autonomous vehicles. (Image courtesy of Cognata.)
Cognata is a lifecycle solution for autonomous vehicles. (Image courtesy of Cognata.)

When it comes to testing the capabilities of autonomous vehicles (AVs), test mileage can be everything. Companies like the Israel-based Cognata understand this fact. Cognata also understands the power that technologies such as artificial intelligence, cloud-based analytics and digital twins can offer the developers of autonomous vehicles. It is technologies like its deeplearning, autonomous and advanced driver-assistance system (ADAS) simulation engine that is allowing developers—who may lack a fleet of thousands of AVs—to speed up the time-to-market of their products.

State-of-the-art deep learning simulation engines are the name of the game. Due to the fact that the simulations can be designed to meet the specific needs of users, a mile in the simulator can represent the same value as hundreds of miles in the real world where events can be few and sparse. Furthermore, they don't require the overhead costs in energy and charging cycles that physical AVs require. The simulator also offers the ability to observe testing at any moment. They are also not inhibited by the lack of 5G infrastructure to relay the state or setting back to analytics.

The Platform and Product

Rapid scene and scenario generation are offered by Cognata’s platform. (Image courtesy of Cognata.)
Rapid scene and scenario generation are offered by Cognata’s platform. (Image courtesy of Cognata.)

Cognata's simulation engine is built around a few foundations: TrueLife 3D Mesh, PhysicsStudio and its simulation engine that combines the two. From those three pillars, the platform offers four layers.

According to Cognata, the first layer is the static layer that "procedurally builds a 'digital twin' of the world." Virtual AVs require virtual assets to interact with, such as the roads themselves, buildings or city infrastructure (like lane markings, traffic lights or utility assets). That is, the static layer of Cognata offers an automated process to generate 3D environments from which specific scenarios can begin to be built to test any particular edge case that a developer wishes.

The second layer is the dynamic layer. This is where assets, such as "drivers, pedestrians and cyclists that replicate behavior customary to the specific geolocation being simulated," are added to the static layer, populating the streets. This second layer builds upon the setting, bringing it alive, so to speak.

Cognata allows for the customization of sensors such as LiDAR and radar. (Image courtesy of Cognata.)
Cognata allows for the customization of sensors such as LiDAR and radar. (Image courtesy of Cognata.)

The third layer is the sensing layer. A feature of Cognata's platform is the ability for developers to integrate sensors into the virtual setting. Default sensors such as popular LiDAR and radar cameras are available, but Cognata's service also allows for the uploading and onboarding of new sensors by users through the leveraging of toolchain technologies.

The final and fourth layer is the cloud layer. This layer utilizes the power and capabilities of powerful cloud computing to deal with the processing burdens of running detailed and involved simulations. The cloud layer allows for the combination of all the previous layers so that complex scenarios can be directly tested in the virtual settings. In a comment to TechTime, the CEO and founder of Cognata, Danny Atsmon said, “According to RAND Corporation an Automomous Vehicle must drive at least 11 billion miles in order to achieve human capability, which defined as 1.9 fatal casualties for 100 million miles. For a standard machine learning training method, you need hundreds of years to reach the goal. But with parallel simulation enabled by the cloud, you can do it quickly, in a practical manner.”

How it looks

As mentioned before, the Cognata platform utilizes its patented TrueLife 3D Mesh, PhysicsStudio and simulation engine to produce convincing and beautiful settings to test AV scenarios.

TrueLife 3D Mesh generates the assets—such as the roads, lane markers, trees, buildings, and pedestrians—in an automated fashion. Users can then edit the scenes to satisfy their edge case’s needs by adding or removing assets. In total, the rendered environments appear realistic and sufficiently detailed. The TrueLife 3D Mesh is a definite challenge to the likes of other AV and ADAS virtual scene generators such as VectorZero. The PhysicsStudio adds in extra features, like lighting and local weather options, that directly affect an AV’s ability to sense its surroundings. Cognata’s simulation engine then brings both these facets together to generate its realistic virtual environments.

Looking to the Future

NVIDIA’s DRIVE Constellation allows for the testing of AVs and ADAS in virtual settings. Cognata and NVIDIA announced a partnership last month. (Image courtesy of NVIDIA.)
NVIDIA’s DRIVE Constellation allows for the testing of AVs and ADAS in virtual settings. Cognata and NVIDIA announced a partnership last month. (Image courtesy of NVIDIA.)

In March 2019, Cognata announced its partnership with NVIDIA to build upon its platform’s foundations and leverage NVIDIA’s DRIVE Constellation to provide bit-accurate scenarios and traffic models to optimize the utility of virtual simulations for the safety of AVs and ADAS.

According to Zvi Greenstein, general manager at NVIDIA, “NVIDIA and Cognata share the vision of using large-scale, cloud-based, open simulation to thoroughly and safely train and test self-driving cars under endless challenging situations. This offering will help accelerate the safe deployment of autonomous vehicles.”

“Highly accurate and scalable traffic model simulation technology is essential to validate autonomous vehicle systems within nearly infinite combinations of real-world scenarios,” added Atsmon.

2019 is an exciting time for AVs, ADAS and digital twin technologies, and it is clear that Cognata’s platform offers an intuitive platform to bring all three sectors together into one.


Recommended For You