How did your take on autonomy and sustainable mobility take shape over your career?
When I started managing the fuel cell vehicle programme at Chrysler in the 1990s, I began to look at the vehicle more holistically than just the powertrain. And that got me thinking: Why shoehorn this fuel cell into a vehicle designed for an internal combustion engine? What if we could design the vehicle around the fuel cell? We could create a lot more design freedom, get more benefits for the customer and make the vehicle far more attractive.
I call this design and technology fusion. That's where the interface of innovation is, between disciplines rather than just inside one discipline. And that thinking led to the "AUTONomy" skateboard concept, which came to fruition at General Motors in 2002.
After that, I married up environmental sustainability and autonomy, to my knowledge for the first time, for the Shanghai World Expo in 2010. I thought, well, even if vehicles are zero emissions and run on renewable energy, there are still externalities. They get stuck in traffic, they collide with each other and with pedestrians, and zero emissions vehicles don’t address those externalities.
This is where we came up with the idea of vehicles that are wirelessly communicating with each other and have sensing on them. Wirelessly communicating means that they can coordinate with each other, because autonomous vehicles by themselves aren't really that coordinated.
Take adaptive cruise control as an example. There is typically a bit of a lag before the sensors respond and you advance behind the vehicle in front of you. Whereas with wireless communications, the vehicles really could move as a train. And you can have much better throughputs, as well as improving safety, because the vehicles can communicate their intent.
So, sensing and communication between vehicles, both combined, would help to address throughput and congestion as well as reduce accidents. This is how the idea of electric vehicles, autonomous vehicles, connected vehicles, which is now common wisdom, came about. We were the first to demonstrate this vision with the Electric Network Vehicles, which people could actually get in and drive at the World Expo.
Where do you see current opportunities for autonomous driving?
First of all, this problem of creating and proving out high-level autonomous vehicles is hard. But there is progress and there are opportunities.
If you took a bird’s-eye view and did a quasi Turing test on vehicles driving in a city, you would find that autonomous vehicles hesitate unnaturally at intersections, for instance, when a human driver wouldn’t. That’s because the AI can’t infer what other road users are about to do, and that is still a challenge for robo-taxis. Besides, driving etiquette changes between the city and the suburbs. Autonomous vehicles are still learning all these behavioural inferences and differences, and it is very hard.
But in highway goods delivery – there may be a business case, even with a driver in the vehicle. The highway is a lot simpler for obvious reasons. You could reduce fuel consumption, you might reduce accidents, and you might be able to reduce insurance, all while improving productivity.
Plus, the trucking industry is facing labour shortages in more countries than one. If you can offer all of the above and a more pleasant work environment, where drivers are monitoring the vehicle 90% of the time and take over only when they need to, for example at the exit or when the weather or road conditions become more challenging, then such a business case might work today and allow you to really prove out autonomous driving with lots of miles of driving. Similarly, automotive manufacturers are thinking of business models where cars operate autonomously on certain well-mapped highways, perhaps as a subscription model.
Could connectivity help?
For robo-taxis it is difficult to see a business case with a driver in the vehicle, but connectivity could help.
My go-to example is hilly San Francisco. If a robo-taxi comes to an intersection, then the vehicle sensors may be looking at the sky and not be able to see any traffic in the crossroad. But if you had cameras at challenging locations inside a geo-fenced area, then they could see what is coming and communicate that information to the robo-taxis. And that knowledge could help improve the confidence level so that they progress through the intersection. They wouldn't hesitate and wait, and the ride would be a lot smoother.
The other scenario is, if the vehicle comes to a stop because it isn’t sure what to do, which is what currently happens. If you had a remote operator connected to the vehicle, who can see what the vehicle sensors are seeing, then the remote operator could send information to the vehicle to allow it to progress through that challenging location. That could work if a remote operator could monitor multiple vehicles because they're not all having problems at the same time, then it becomes cost effective to do that.
People talk about infrastructure being expensive, but if you're doing it in a geo-fenced area where the robo-taxis are operating, then it's a confined problem. It may be a good investment for robo-taxi companies because it helps their system operate more seamlessly and effectively.
So, connectivity can really help.
So where might we first see driverless vehicles in urban environments?
I think there's a real possibility that China could take the lead, because they are looking at all this more pragmatically and embracing this fusion of connectivity and automation. I think that some of the companies in the West are thinking of this as a software research problem, and they don’t want to rely on what they regard as a crutch that might level the playing field between them and their competitors. But that’s a blind spot. In China, they could be introducing autonomous vehicles sooner because of the additional help that vehicles get from urban infrastructure. Of course, there are privacy issues associated with putting up cameras. But you could address those by blacking out people's faces, or maybe use lidar instead.
Tell us about your passion, progress with Afreecar.
I have done a lot of volunteer work in Africa, and that’s where the name Afreecar comes from.
The idea for Afreecar started as developing a solar-powered vehicle for Africa that could also provide electric power for purposes such as lighting, charging phones and so forth. But it could also be used for transport.
And then over time my thinking evolved and I thought more around a kit that could be used, because I don't want to prejudge what the right vehicle is when everybody may have their own non-motorized vehicle already – anything from a wheelbarrow to an oxcart, to a bicycle, to a wheelchair.
So I'm developing a kit, the size of a briefcase that contains lithium ion batteries and motors and solar panels, which would then be put on the roof of any non-motorised vehicle. The idea would be that solar panel would charge the batteries. And the batteries would power the electric motor that would cause this non-motorized vehicle to be able to move uphill and carry more stuff. And at the same time you could plug in anything from cell phones to water pumps. So, instead of people having to buy a vehicle, people just buy this kit that attaches to their vehicle. And it not only provides a transport power assist, but also allows it to become a mobile power source.
By the way, such a solution can work in Europe or North America too, in agriculture or healthcare, perhaps without the solar panels. Think about a hospital, nursing home or hospice, with beds, wheelchairs, gurneys, meal carts, linen carts. These are all non-motorized vehicles.
If you had a kit that could be shared by all of these different types of vehicles, instead of every single one of them needing power assist, you could have a shared power assist solution that could retrofit to all of them. That would be a lot cheaper, and you might reduce your need for labour, workplace injuries and even transmission of diseases between caregivers and patients, because now you don't need three or four people to move a hospital bed, you only need one worker with a power assisted vehicle.
So this idea of a kit that is universal that I'm developing. We are testing it now here in Chicagoland. Afreecar’s vision is to send it to Africa afterwards.
To learn more about Afreecar, and its mission of Providing Sustainable and Affordable Mobility and Power for All the World's People, visit afreecar.org.
Thank you, Chris, for speaking with the Autonomous Driving Tech team at TTP.
The sensors for autonomous vehicles are getting smarter and more accurate, enabling vehicle developers to reach high levels of autonomy in prototype systems. The next challenge is to reduce the cost of manufacture and integration to a point where wider adoption is possible. TTP’s Autonomous Tech team are developing the enabling technologies required for advanced autonomous vehicles.