News - General News Technology
Automated cars won't stop crashes: Sweatman
Crash land: The former head of the University of Michigan Transport Research Institute Dr Peter Sweatman said cars will still crash even when autonomy is everywhere.
Mixing humans and driverless cars will be tricky, says autonomous expert
5 February 2016
BEING an occupant in a driverless car will not guarantee that it will not be
involved in a crash, according to a leader in automated and connected vehicles.
Dr Peter Sweatman who designed and ran several large trials of automated and
connected vehicles while he was head of the University of Michigan Transport
Research Institute, said crashes will be difficult to avoid even when a large
proportion of the vehicles on the road are automated.
The authorities and consumers will have it in their power to choose a level of
safety, Dr Sweatman told an Intelligent Transport Systems Australia meeting in
“So, what’s acceptable safety? It’s an increment better than what you have at
the moment, but when you go to a totally driverless situation, you can choose
what level of safety you want. How are we going to do that?
“Realistically, there are still going to be crashes, and that’s going to be
controversial if driverless vehicles are crashing,” he said.
“We need to get beyond that and realise that actually the safety is going to be
that much better.”
Left: Former head of transport research institute
University of Michigan Dr Peter Sweatman.
Dr Sweatman said that while automated cars will generally do what they are
programmed to do, humans will continue to do what they want.
“One of the problems with this, and we have seen it in Mountain View with the
Google testing, is that machines do exactly what they are programmed to. (Which
is why they are automated, not autonomous.)
“So if those machines are programmed by a responsible company to stop on a red
light or an amber light, then that’s what they do, every time.
“And, of course, they’re in a traffic stream with human-driven vehicles and the
human is looking at the signals and trying to cheat, you know. The two don’t go
together too well,” he said.
“So the challenge there is whether to get very quickly to a fully automated
system so that it’s not an issue, or to get real about what the standard really
needs to be for both the machine and the human.
“At the moment, the machine is programmed conservatively by a responsible
company and the human is doing what they always do.”
According to Dr Sweatman, that is the reason there will continue to be crashes,
at least until there is a high proportion of automated vehicles on the road,
but the crashes will be less severe.
“The crashes that have been experienced by automated vehicles have been of very
“So, we’re not going to be in a zero-crash situation. There will always be some
situation that isn’t being dealt with by the machine.”
Dr Sweatman said governments and consumers will still have a choice about how
much safety they want even when there is a high proportion of automated
“For example, if you are designing something where the vehicles are habitually
a lot closer together, for efficiency reasons, then the closer together those
vehicles are, the more errors there are going to be in the system.
“Even from that perspective, there is a choice being made. Our choices at the
moment aren’t articulated,” he said.
“There is always a choice in this, whether or not it is expressed. That will
still be the case, but the serious crash rate, the fatality rate, will be a
factor of 10 less.”
Dr Sweatman left the University of Michigan at the start of the year and has
formed a partnership, CAVita, with Abbas Mohaddes, the former chief executive
of Iteris, a company that sells traffic management systems and weather analysis
to the agricultural sector.
CAVita will act as an “online dating agency” for all the public and private
entities trying to understand what roles they can play in the deployment of
automated and connected vehicles, Dr Sweatman said.
He said there are now thousands of trials in these two areas being conducted
across the US and a huge diversity of technologies being developed.
“How do they deal with this diversity issue? That’s why we set up CAVita. But
we are interested not only in the connected vehicle and the automated vehicle,
but other technologies are important too, whether that’s the Internet of
Things, Big Data, and this whole Smart Cities movement.
“We want to keep ahead of the curve as to who the important players are, who
the most important partners are and making sure that the government agencies
are well-informed about this, too.”
Dr Sweatman said every state in the US will have to develop policies for these
technologies and they will have to invest in infrastructure to facilitate their
“It’s all about managing risk and if they feel the technology they deploy this
year will be defunct in a couple of year’s time, the automotive industry will
have moved on in some way and leave them behind. There are these fears and we
need to be able to manage all of that and they need help to do that.”
Before he went to America in 2004, Dr Sweatman had played a prominent role in
road research in Australia. In 1976, he established the heavy vehicle group
within the Australian Road Research Board (ARRB) and pioneered the scientific
analysis of commercial-vehicle operation on the highway system.