Visteon's Advanced Automatic Driving Verification and Validation Measures
Shanghai, China, June 27, 2019 - At the 2nd Annaul China Automated Driving Test Validation Technology Innovation Conference, ADAS/Automated Driving Software Project Manager from Visteon Technology & Service R&D Center, Ms. Kavita Kshirsage shared Visteon's ADAS & Auto Keynote speech on
Verification & Validation Approaches for Highly Automated Driving (HAD)
Ms. Kavita first introduced Visteon's autonomous driving platform "DriverCore". Last year, Visteon brought a concept of DriveCore, and this year Visteon brought more and more content. Modular, customizable hardware computing platform; high-performance, real-time communication of applications and algorithms via in-vehicle middleware; API interface and visual interface, all of which are flexible features of DriveCore. The emergence of DriveCore gives the industry a new way of thinking, providing a new variable in the autonomous driving industry chain, linking AI companies and auto companies to provide a LEGO idea.
Next, Ms. Kavita elaborated on Visteon's ADAS & Autopilot Test Verification Method & Acceptance Measures. The advanced driver assistance system is an automated system that improves safety and improves the driving experience. ADAS is more complex and powerful, and the number of use cases required for testing has increased significantly, and the focus has shifted to analyzing and synthesizing a large number of test scenarios with reasonable effort. ADAS / AD validation includes a set of use cases that are prepared to test the strategy and derive from the requirements (functional and non-functional) that the system needs to meet. The space of possible scenarios spans multiple dimensions: road geometry, behavior of drivers and other traffic participants, weather conditions, vehicle characteristics and vehicle variants, spontaneous component failures, event scheduling, and the like. While road testing can ensure the durability of the system, it may not include testing feature behavior for dynamic scenes because it is difficult to create them. Therefore, it is widely believed that true test drives must be extensively complemented by virtual test drives. Compared to vehicle testing, a virtual test domain is characterized by a deployment model to simulate the functionality or components of a vehicle that is not yet available.
Ms. Kavita also elaborated on the criteria for accepting Level 3 functions and the standard for full V-cycle deployment. Finally, Ms. Kavita used the Highway Pilot as a case to illustrate the scenario-based advanced autopilot test method and demonstrated the Visteon Driver-in-Loop test.