To physically verify the safety of an autonomous vehicle (AV), Toyota estimates that a vehicle would have to undergo tests that ran along about 8.8 billion miles. Of course, testing on this many miles of test tracks is pretty much impossible. Automakers must rely on a significant increase in virtual testing and “mixed virtual-physical testing,” along with increased physical testing
When it comes to being able to mass produce autonomous vehicles, it’s not just about developing prototyping software. It’s about production electronics. It’s about having full validation and verification capabilities, both virtually and physically, of the vehicle itself. It also includes systems-driven product development. Automakers need to make sure there’s design data continuity not only for developers, but throughout the supply chain.
How will all of these parts come together so vehicle manufacturers can begin to mass produce autonomous vehicles?
To that end, a number of partnership have sprang up within the past decade to support AV vehicle testing. For instance, Siemens PLM and the American Center for Mobility have combined forces to enhance the ability of automakers to meet the validation and verification challenges.
The partners say that challenges include:
Designing & developing robust automotive-grade electronics. While some electronic systems in a modern vehicle are safety-critical, such as ABS braking controller, many aren’t, such as the car’s radio. As we move into the self-driving cars era, the number of safety-critical electronic systems increases exponentially. For consumers to accept autonomous vehicles, and for automotive manufacturers to want to sell them, electronics must react in milliseconds and be 100 percent reliable under all operating conditions.
Designing, developing and integrating sensors. Autonomous vehicles are essentially conventional vehicles – but with a huge increase in sensors and software. These sensors must be both low-cost and hugely reliable. The sheer number and the wide range of required sensors is challenging. Automakers must incorporate cameras, radar and LIDAR systems, and more importantly, must fuse all information from these sensors so the vehicle can properly identify what’s happening around it – and put it all into a single message to identify the real threats on the road.
Developing robust software and controls. Software is the at the heart of the The complexities involved in creating autonomous vehicles that can properly sense the environment, decide what must be done and then act on those decisions, are daunting. This sense-decide-act cycle needs to happen in milliseconds and must be better than humans. The software and controls must also keep learning and evolving over time so these vehicles become better and better over time.
Partnering and close collaboration. Automotive design and development has always been a collaborative effort with a vast supply base, with more than 70 percent of a vehicle typically coming from suppliers. With autonomous vehicles, this communication and close collaboration is amplified. Not only are automakers relying on suppliers for more high-end systems, but they’re collaborating with a wide range of non-traditional partners. Automakers are collaborating with global high-tech firms, such as Google, all the way down to small start-up firms. An even greater degree of collaboration is required to ensure that 100 percent of requirements are understood across the partner system, and that all partners flawlessly deliver on those requirements.
Validation and verification. Automakers currently perform a huge amount of virtual and physical testing to ensure their vehicles safely meet the needs of a wide range of customers. In a traditional vehicle, the driver is always present to react to the infinite number of real-world driving challenges that we face. In an autonomous vehicle, the vehicle must take on the role of thinking for the driver – in addition to meeting all the usual requirements. This is daunting!
Each of these challenges must be studied in more detail, of course.
Currently, however, partnerships between private and public industry, between automakers and engineering software makers, between universities and private industry are a big step toward meeting those challenges.
Meanwhile, Siemens offers itself as a partner for vehicle manufacturers looking to mass produce autonomous vehicles. It’s recent partnership with the American Center for Mobility will help us prepare our customers for the changes coming to automotive manufacturing, according to Siemens.