Why Digital Twin could be the secret to unlock autonomous driving, safely
While the tragedy earlier this year in Tempe, Arizona involving an autonomous vehicle colliding with a pedestrian was met with significant media attention and public outcry, it is unfortunately not the first major accident involving a driverless car. In 2016, a Tesla model-S in semi-autopilot mode was involved in a crash with a tractor-trailer in Newark, Delaware.
The number of accidents involving driverless cars remains incredibly small compared to the staggering number of road accidents that happen every year, but these incidents are naturally attracting a lot of attention. Legal proceedings in the 2016 accident arrived at the decision that Tesla was not liable as there was no defect in the autopilot system. In fact, it was the responsibility of the driver in the vehicle to be alert enough to regain control where necessary.
Nevertheless, as companies continue development and innovation in the field of autonomous driving, ensuring safety is the primary consideration. Pressure will be on manufacturers and technology companies to make sure that development and testing of cars is managed safely, with limited or no damage to individuals and property. For governments, the challenge will be to balance driving (no pun intended) the development of a technology that has the potential to dramatically improve society, alongside the critical interest of public safety.
The primary technological challenge with developing autonomous vehicles, is preparing for the infinite number of potential scenarios that can play out on the road. Variables such as environmental conditions, other vehicles, pedestrians and traffic signs means there is such a high number of potential combinations that it is physically impossible for these to all play out during testing on the roads. Additionally, the current model of using people to test autonomous vehicles means relying on imperfect humans to monitor imperfect technology and respond to emergencies when they arise.
Thankfully, a solution exists. The digital twin is a sensor-enabled digital model of a physical object that simulates the object in a live setting. In effect, this virtual model enables companies to have a complete digital footprint of their product from design and development through to the end of the product life cycle. Digital twin is crucial for the development of autonomous vehicles as it eradicates the problem of having to put a driverless car through an infinite number of situations to perfect the algorithm. These scenarios can now play out in the virtual world, rather than needing to be learned physically.
Additionally, as more product liability cases go to court – and I suspect that there will be many to come – the digital twin will help companies prove in proceedings how technology was used to test and verify safety processes.
Digital twins are already used to simulate specific complex deployed assets such as jet engines and large mining trucks. The technology enables monitoring and evaluation of wear and tear and specific kinds of stress as the asset is used in the field. In fact, the technology has already had a high success rate in the aerospace industry where it was used to develop the autopilot function and is still used for system safety and analyses.
In addition to improving safety in the development and testing of driverless cars, digital twin will also increase the speed to market, reduce defects during manufacture and crucially, it will support emerging business models. Car manufacturers will essentially become software vendors by using the digital twin to provide drivers with software updates and maintenance recommendations. This will drive entirely new revenue streams.
If they really want to realize the full potential of autonomous vehicles (and hey, let’s dream for a minute, maybe flying cars in the future as well), automotive manufacturers will have to embrace the digital twin. In the meantime, it needs to play a key role in ensuring autonomous vehicles are as safe as possible.
Pete Carrier