About the author: Ashwin Ballambettu Pai is an Associate Director at Zinnov with the Global Engineering Service Provider Practice.
Self-driving Technologies have come a long way from hands-free and minds-off driving to driverless pods in a short span. Luxury car maker BMW is expected to launch a fully self-driving car by 2021¹. 10 million self-driving cars would potentially be on the road by 2020². Arguably, the automotive industry has its priority on self-driving technologies for the future.
According to Zinnov, global R&D spend on self-driving technologies in 2015 was close to $6B. Toyota has earmarked $1B over next 5 years to develop advanced driver assistance systems³. Technology giant Google is spending $30M YoY on autonomous cars⁴. Tier 1 supplier Bosch is leveraging artificial intelligence(AI) to develop its autonomous technology⁵. Uber has started testing its own self-driving car in Pittsburg, PA⁶. In summary, we would potentially see plethora of autonomous technology offerings from multitude of players – OEMs, Tier 1 Suppliers, Technology Giants, Ride sharing companies, and more.
Thanks to hyper competition self-driving technology is advancing and maturing at rapid pace. However, a recent fatality involving a Tesla Model S electric sedan using the Autopilot feature has casted doubt on the safety aspect of autonomous driving technology⁷. Earlier this year, Google’s self-driving car caused an accident with a public transit bus⁸. Mid last year test cars from Google and Delphi Automotive nearly collided, a first such instance involving a pair of self-driving cars⁹.
Autopilot systems comprise of disparate hardware and software sub-systems such as image and proximity sensors, Radar, Lidar, vision processors, AI algorithms, etc. Also, such sub-systems are developed/supplied by multiple industry players adding to the complexity and reliability of autopilot systems. Ironically, Tesla and Mobileye – a provider of Autopilot technology to the former, are in disagreement on why the Autopilot did not apply brakes¹⁰.
A deeper inspection into the above incidents point at gaps in testing of self-driving technologies. To achieve 100% test coverage for countably infinite number of potential scenarios that could lead to an accident is humanly impossible. Also, testing conditions exacerbate due to interoperability challenges that could potentially arise between self-driving technologies from multiple players. Further, in countries such as India with no lane discipline, future autopilot systems must also deal with chaotic traffic including pedestrians as well as holy bovines. Having said, 1.23 million vehicle collisions occurred with deer in the US during the 12-month period ending June 2012¹¹.
(Image source: Google Self Driving car)
With too many corner case scenarios that would potentially escape test, automotive industry is bound to face regulatory hurdles further delaying wide spread adoption of self-driving cars. Regulatory bodies would ensure parity for safety standards across self-driving technologies irrespective of premium or mass market OEMs. Automotive players should seize the moment to collectively address the challenge and win.
First, the automotive ecosystem including the regulatory bodies should jointly formulate the policies and regulations around self-driving cars. Second, there should be an industry wide effort to define and adopt open standards/ technologies based platform for self-driving systems. This would ensure transparency about the limitations of such autopilot systems. Finally, define a robust framework for system test – modeling, simulation, test and validation. Also, partner with extended ecosystem such as engineering services players to accelerate test and ensure zero test escape.
“Our vision is that by 2020 no one should be killed or injured in a new Volvo car.” – Håkan Samuelsson, President and CEO, Volvo Cars
In conclusion, it is the spirit of competition that led to the advancement of self-driving technologies. However, the primary challenge of automotive industry is to ensure passenger safety which can perhaps be achieved through greater collaboration.
References:
- https://techcrunch.com/2016/07/01/bmw-mobileye-and-intel-are-building-a-full-self-driving-car-for-2021/
- http://www.businessinsider.com/report-10-million-self-driving-cars-will-be-on-the-road-by-2020-2015-5-6?IR=T
- http://www.businessinsider.com/toyota-is-spending-1-billion-on-advanced-safety-features-2016-6?IR=T
- http://www.oesa.org/Publications/OESA-News/November-2015/Googles-Self-Driving-Car-Strategy-and-Implications.html
- http://www.bosch.be/en/be/newsroom_16/news_15/news-detail-page_73152.php
- https://techcrunch.com/2016/05/19/uber-confirms-its-testing-self-driving-cars-in-pittsburgh
- http://www.reuters.com/article/us-tesla-autopilot-idUSKCN0ZH4VO
- https://www.wired.com/2016/02/googles-self-driving-car-may-caused-first-crash/
- http://fortune.com/2015/06/26/google-self-driving-lexus-audi/
- http://electrek.co/2016/07/01/tesla-autopilot-mobileye-fatal-crash-comment/
- http://www.automotive-fleet.com/news/story/2012/10/state-farm-deer-related-collisions-up-7-7-in-past-year.aspx
