Using simulation to drive Toyota’s automated driving system development process

About TRI-AD

Toyota Research Institute-Advanced Development, Inc. (TRI-AD) develops innovative software that will enable a vision of Toyota Motor Corporation (Toyota), “Mobility for All.” TRI-AD's mission is to create new technology and advanced safety systems for the world and oversee Toyota’s newly announced Woven City, which will serve as an incubator for smart city design, connected mobility, and robotics technology from Toyota and its partners.

2018
580
Founded
Office
Tokyo
Employees (as of September 2020)

"Applied Intuition works hand in hand with our engineering teams to integrate their tools with our quite complex codebase.

The depth of their knowledge coupled with their agility and the rich feature set of their products help move our project quickly and successfully. Applied's responsiveness to our needs have helped us to accelerate."

Nikos Michalakis
VP of Software Platform, TRI-AD

Synthetically created expressway for perception system testing (1)

Goals

TRI-AD is at the forefront of automated driving vehicle development, incorporating the latest trends and technology into their development process. They focus on developing algorithms that aim to deliver the world’s most secure, comfortable, and above all, the safest mobility experience.

To accelerate the deployment of automated driving experience in Toyota’s models, the team is focusing on the following key areas.

01
Automated driving control algorithm development: Develop algorithms capable of safely navigating through traffic-heavy highways
02
Safety and functional safety testing: Ensure compliance with ISO 26262 etc., and system/component failures are properly addressed
03
Expanding coverage to test edge cases: Verify their perception module in various synthetic 3D environments that are difficult to reproduce on expressways and other motor-vehicle-only roadways
04
A scalable approach: Test tens of thousands of scenarios concurrently to speed up the development cycle

"End-to-end validation of a vast array of real-world scenarios is critical to ensure the safety of our customers.

Our automated driving software is increasingly more complex and we need powerful and scalable simulation environments to efficiently and comprehensively verify the system.

Applied Intuition has provided us with an easy-to-use tool and we will continue to rely on them to speed up our development cycle."

Kenichiro Aoki
Staff Engineer, Teammate Features Development Team, TRI-AD

Synthetically created expressway for perception system testing (2)

Approach

TRI-AD has partnered with Applied to test its self-driving system end-to-end under a vast number of scenarios quickly, safely, and comprehensively.

Automotive-grade simulation engine: Applied’s software is used in TRI-AD's extensive testing process and informs TRI-AD's development decisions
End-to-end system testing: Applied’s simulation tools enable end-to-end verification of complex systems including the perception module. CI environments of such systems are also created
Seamless integration: Integration was customized to meet TRI-AD’s requirements (e.g. support for vehicle dynamics model integration)
Large-scale simulations in the cloud: An environment to test large-scale scenarios in the cloud in a parallel state was established
Evaluation under a wide range of edge cases: Synthetic data enables the scalable generation and verification of situations by changing parameters such as weather, fog, time of day, and visibility
Testing the perception module using synthetic data: Sensors such as Toyota’s proprietary cameras are modeled accurately; virtual 3D environments specific to TRI-AD’s ODDs are generated procedurally to test TRI-AD’s perception stack without driving on roads

"Applied Intuition's tools combine the ease of scenario creation and scalability. They also have the flexibility to add features such as those for sensor model implementation.

We feel supported by Applied team's proactiveness in addressing our technically challenging requests."

Tadashi Okubo
Senior Engineer, Teammate Features Development Team, TRI-AD

Impact

01
TRI-AD has gained abilities to build software in a virtual environment and test their system end-to-end, leading to a shorter development cycle.
02
TRI-AD can now develop high-quality software without re-work as a result of focusing on the CI environment and creating an environment that tests software regressions constantly.
03
Comprehensive tests are completed more quickly because simulations have scaled their AV verification process.

Get stories from autonomy engineers