Revealing “Invisible Risks!　Prediction Algorithm Validation in the AI Era Automatic Generation of Blind Spot Scenarios

Although the Safety Evaluation Framework (Ver. 3.0) has already addressed traffic disturbance scenarios, it has not yet been able to address recognition disturbance scenarios, has it?
Especially for blind spot scenarios, it is difficult to create and evaluate scenarios, and the verification method may be unclear.

In order to realize “may drive” by AI, it is important to verify the safety of AV/ADAS using blind spot scenarios.
We will present a method to automate the creation of blind spot scenarios and visualize the coverage rate by quantifying it. In addition, we will provide hints for speedy response to JAMA’s safety assessment framework (Ver. 3.0).

The complex and time-consuming task of creating blind spot scenarios will be greatly shortened, and the coverage rate of the test will be intuitively grasped numerically and graphically by Coverage.
You can enhance the testing of blind spot area prediction algorithm validation and accelerate compliance with the JAMA Safety Assessment Framework (Ver. 3.0).

Testing methods for “might-be driving” by AI can accelerate the response to blind spot scenarios.
If you want to reduce the time and effort required to create blind spot scenarios, or if you want to realize safety evaluation for blind spot scenarios, please check this page.