Artificial intelligence (AI) and the modern vehicle go hand-in-hand when addressing safety, advanced driver-assistance systems (ADAS), and the move toward self-driving cars. Designing these systems requires lots of testing, and the only practical method to do most of this is simulation. Real-world testing is a requirement, too, but simulation is more effective in providing different, controllable scenarios. It’s also less expensive and faster so that more variables can be changed and controlled.
Working AI into both the simulation support as well as the AI used in the application is becoming more complex as more machine-learning (ML) models are being deployed. I talked with Seth DeLand, Product Manager for AI products at MathWorks, about the challenges of using simulation and AI when developing automotive applications.
Developers are going to require tools to handle the massive amounts of data as well as the large number of complex simulations to provide useful information about the quality of the system, let alone assisting in bug identification and system optimization. It’s more demanding than other applications of simulation and AI, as the results often need to meet safety requirements such as ASIL D and ISO 26262.
This is the third video in the series. The others include:
- Innovation vs Obstacles: How to Integrate AI and Simulation
- Innovation vs Obstacles: AI Models and the Role of Simulation in the Auto Industry
- Innovation vs Obstacles: The Future of AI for Simulation in Automotive Applications (video above)
