Simulation will be a cornerstone technology for autonomous vehicle implementation, underpinning system training, testing and validation, according to a report from ABI Research. While being an asset to all autonomous technology developers, simulation will prove a “must-have” for adopters of deep-learning approaches, bringing some much-needed validation and confidence before deployment.
“Almost every OEM has positioned connected and autonomous vehicles front and center of their strategy for the next decade, lauding the advantages for safety and efficiency. However, demonstrating these advantages will require OEMs to accrue an extraordinary number of autonomously driven miles,” says James Hodgson, senior analyst at ABI Research. “Simulation will not only build up the volume of the driverless experience but will also enrich its variety–allowing OEMs and other autonomous technology developers to test their systems in rare and potentially dangerous scenarios without risk of damage or loss of life."
Accruing the tens of billions of miles necessary to have confidence in autonomous systems would require the deployment of at least 3 million unproven autonomous vehicles over the course of 10 years. Waymo, already leading the market in terms of autonomously driven miles, has made effective use of simulation through their Carcraft initiative, covering in one day of simulations as many autonomous miles as they did between 2009 and 2018 with their physical prototypes.
Now multiple simulation solutions have come to market, enabling OEMs to test and validate their autonomous systems before their self-imposed deadlines arrive within the next three to five years. Some vendors, such as NVIDIA and AIMotive will feature simulation as part of their broader autonomous driving solutions, while specialized vendors such as Cognata and Metamoto are helping to ease the integration of simulation into autonomous vehicle development workflows through a simulation-as-a-service offering.
“There is a growing consensus on the components that will be needed to enable autonomous driving. What OEMs really need now are technologies and partners that can support them in addressing the remaining pain points that are holding back implementation. Simulation is set to be an important technology in giving autonomous system developers confidence in their systems before deployment,” Hodgson concludes.