Sierhuis is working on a system he calls Seamless Autonomous Mobility (SAM) which will combine autonomous vehicles with a back-to-base operator, connected via cloud computing, that will be able to help the car when it gets stuck in an unpredictable real-world situation. The initial plan for the system is to use it as an autonomous taxi or delivery service, with either Nissan or the fleet operators paying for a human “mobility manager” to assist the cars. At first that mobility manager would look after up to 50 cars but that could be scaled up over time as the car’s artificial intelligence (AI) learn more about navigating obstacles in the real world.
While his ideas may run counter to many in the industry, Sierhuis has an impressive resume. He joined Nissan from NASA, where he worked on AI for the International Space Station and Mars-bound robots. From his learnings there he believes what he calls a human-in-the-loop systems is a must for future autonomous cars. “People have this view that autonomy has to be completely without any interaction making a decision with humans,” Sierhuis told media at the Nissan Futures conference in Hong Kong last weekend. “But we as humans don’t even do that. I think it’s a natural way, I think everybody is working on it in some way.”
He admitted his ideas are radical within the industry but he is adamant that having human interaction is the only way forward given the unpredictability of driving, where drivers break road rules and can act randomly.
“In the beginning people said I was crazy, ‘What are you talking about? You’re here to develop autonomous systems,’” he explained.
“Then you show them examples and say ‘What are you going to do in this case? Are you going to allow it to break rules? If you’re […]