Notes from #WebSummit: Autonomous driving & handling the handover
A discussion on the ethical dilemma of humans vs. machines controlling the wheel.
Details
Date: November 7, 2017
Time: 11:20
Conference stream: AutoTech / TalkRobot
Speakers
Zach Barasz, Partner, BMW i Ventures
Adam Kell, Partner, Comet Labs
Karen Francis, Board Director, Nauto
Jim Motavalli, Auto Expert and Journalist, NPR/National Public Radio
Web Summit Summary
Before driverless cars go mainstream, there will be an in-between period in which human and machine will alternate control of the wheel. An ethical dilemma too complex for the computers may mean the rider suddenly has to take over. What’s the best way to handle this transition? The board director of self-driving startup Nauto, a partner from BMW i Ventures and an AI investor weigh in.
Main Theme
As we’re designing technology to make vehicles work seamlessly without humans, one of the biggest debates is when we should give up total control of the road. This panel discussed the different options we have, as well as key ways to aid the transition to enable human drivers to be ready for the hand over.
Key Points
Everyone has a different take on when we should give up total control to our vehicles. Some companies are pushing for innovation and claim as early as 2019 we’ll see vehicles driving solo on the road. Others say we have to carefully transition so drivers get used to the concept of the take over. Here are the key things to keep in mind as we design solutions for this transition:
Humans just can’t be trusted to respond in time. Giving signals to the driver is important but people are horrible at paying attention in vehicles. The last thing you want is for someone to take over and they are too distracted to make an extremely important split second decision.
Think of what’s better for the piece of mind of the driver. We need to work on finding a way to identify when people aren’t paying attention to warn them in the correct ways. We need to take advantage of tools like GEO fencing, haptic feedback, etc. to give cues and as much advance warning as possible.
Handling every use case will be the largest hurdle. The biggest problem with handling the handover is the sheer amount of edge cases. There are hundreds, if not thousand of combinations that can happen in any situations, such as a construction zone that make achieving Level 5 autonomy so difficult.
This process will be a gradual, not sudden transition. We need to focus on the most problematic issues first while we accelerate autonomy. There will be a transition over time so we can continually learn from it, not dissimilar from when cars and horses shared the roads for 30 years.
Reflections
The biggest take away for me was although this user problem is more extreme than the problems I solve day-to-day, the solutions tie back to the main principles for solving design problems: focus on the core user problem first and continue iterating until the lesser use cases can be fixed.
These notes are brought to you by TWG: software makers to the world’s innovators.
Want to know what sessions we’ll be at next? Our chatbot pal Web Summelier can hook you up. Want to receive a summary of all our notes when Web Summit’s over? Sign up here.
Psst! We’ve published many more insights on technology, design, and all things software on our blog, The Almanac.
