Artificial intelligence (AI) and machine learning are becoming increasingly important as the automotive industry explores ways to develop autonomous car technology. Should we fear the rise of the machines? Or celebrate the imminent demise of poor quality human decision making?
In public road testing, every crash involving autonomous cars has been down to human error – until the recent Google car / bus incident. The Valentine’s Day incident involved, according to a report filed with California’s DMV, a Google car pulling into a lane occupied by a bus. The resultant low speed collision (Google AV: 2mph, bus: 15mph) had been anticipated by the ‘driver’ of the Google AV, which was in autonomous mode, but because the ‘driver’ believed “the bus would stop or slow to allow the Google AV to continue” around the obstacle in its lane, there was no human intervention.
While state and national regulators debate with organisations like the EU and NHTSA about whether a computer can be classed as a driver the development of machine learning and AI advances
Machine learning in action? Google says it has since simulated the incident in thousands of variations and refined its software. “From now on, our cars will more deeply understand that buses (and other large vehicles) are less likely to yield to us than other types of vehicles.”
It is for others to decide whether, in Google AV versus bus, man or machine was at fault; either way, while state and national regulators debate with organisations like the EU and NHTSA about whether a computer can be classed as a driver – and thus whether driverless cars require manual controls and, ironically, drivers – the development of machine learning and AI advances.
Hyundai and Kia, the Renault-Nissan Alliance, Daimler… numerous OEMs and suppliers have committed to bringing autonomous drive technology to market. Toyota has reportedly recently acquired in full Jaybridge Robotics, a 16-strong autonomous-vehicle company based in Massachusetts – just down the road from one of its two Toyota Research Institute (TRI) facilities (the other is in Palo Alto, California). With mining and military at the forefront of autonomous vehicle development, it should come as no surprise that Toyota might be interested in a company that specialises in “complete, turnkey solutions for a range of applications in the agricultural, aggregates and mining industries” and considers the US Army and US Navy among its technology and business partners.
With a million road traffic deaths annually caused by human error, can we afford to let this fear extend to AI in automotive applications?
TRI focuses on the application of AI and advanced mobility solutions. Although TRI Chief Executive, Dr Gill Pratt, sees the current level of autonomous car development as being well short “of the finish line”, he sees significant potential for machine learning, AI and robotics. At CES 2016, he said that “robots could become for today’s Toyota what the car industry was back when Toyota was making looms.”
How far will machine learning and AI go? Professor Stephen Hawking says AI could threaten our very existence, and that we should be fearful of it. Should we? And with a million road traffic deaths annually caused by human error, can we afford to let this fear extend to AI in automotive applications?
Martin Kahl, Automotive World’s editor, will be chairing a panel discussion at Autonomous Car Detroit on 16 March 2016, featuring Ken Laberteaux, Senior Principal Scientist, Toyota Research Institute; John Maddox, Assistant Director, University of Michigan Mobility Transformation Center; Priyantha Mudalige, R&D Group Manager, General Motors; Kirk T. Steudle, Director, Michigan DOT; and Steve Underwood, Director – Connected Vehicle Proving Center (CVPC).
For more information, go to: autonomouscardetroit.com
