What if a robot drove your car?
In the 1980s the global TV hit Knight Rider not only made a star of David Hasselhoff, it also made the idea of a car driven and controlled by a robot part of popular culture. Now the concept is moving from the TV screen onto the roads of Silicon Valley.
"The steering is entirely drive-by-wire, it's computer controlled," says Holly as we hurtle around the campus of Stanford University in a weird and wonderful contraption which would not seem out of place in a Mad Max movie.
"It's all electronic - there's no mechanical connection so we can programme it to do pretty much what we want."
Holly works at Stanford's Center for Automotive Research, based in Palo Alto, in the heart of California's Silicon Valley.
The machine she's driving is known as the X1. It looks like a cross between a car and a beach buggy, with half an electronics laboratory thrown in.
There's no bodywork, just a frame made out of metal tubes. We sit in bucket seats, held in by race harnesses.
In front, you can see every twitch of the chunky red suspension springs as the large alloy wheels bounce over Stanford's many potholes.
The whole thing is festooned with wiring, switches and important-looking electronic gadgets.
The X1 is a test-bed for the kind of technology which might seem exotic today, but which may eventually be commonplace.
It's one of a number of vehicles created by the organisation which is on the front line of research into electronic systems that are designed initially to help human drivers, but may ultimately replace them altogether.
Many of our cars are increasingly driven by artificial intelligence - they have engine management systems, computers which are designed to make sure the engine runs as efficiently as possible.
ABS detects when a driver has used the brakes too violently and prevents the wheels from locking up, and traction control can stop the wheels from spinning uselessly when the driver accelerates in slippery conditions.
For the moment, the driver remains in charge, but research into driverless cars or "vehicle autonomy" as it's known in the industry, is already well advanced.
Internet giant Google has actually been operating a fleet of self-driving cars on the streets of California for some time, but other multinationals such as Volkswagen, Bosch, General Motors and Mercedes are also working on autonomous systems.
As a result, Silicon Valley is increasingly becoming a new hub for automotive engineering.
In February, Japanese carmaker Nissan became the latest manufacturer to set up a research office in the region, specifically to focus on developing self-driving technologies.
But why do we actually need driverless cars? According to the executive director of the Center for Automotive Research at Stanford, Sven Beiker, a key aspect is safety.
"We're very interested in what the driver is good at, and what the computer is better at doing, and we're working on the handover between the driver and the vehicle," he says.
"So what we're looking at is when the driver gets him or herself into trouble, as often happens when driving too fast, the vehicle could take over."
But safety is just one aspect. Driverless vehicles could also become a vital tool for cutting congestion.
"If you look at a highway, even at rush hour times, only about 20-25% of the surface area is actually occupied by vehicles," says Mr Beiker.
"We need a lot of space around the car in order to manoeuvre but if you have a lot of top-notch sensors and control systems, you can probably get those vehicles much closer together."
If those cars could communicate with one another as well they would also be able to identify traffic hotspots and avoid them.
But there is another way in which self-driving vehicles could cut congestion. That would be to make it easier for people to rent vehicles when they really need them - rather than owning a car they seldom use.
'Waste of space'
"Privately owned cars are a big problem," says Prof Kent Larson, head of the Future Cities programme at the Massachusetts Institute of Technology (MIT) in Boston.
"Most cars in cities aren't being used 90% of the time," he says. "The rest of the time, they're parked up and wasting valuable space. But if we can move to a shared model, we can increase the use of a single vehicle by at least five times."
That's where driverless cars come in.
"The Holy Grail is a vehicle that can drive itself and park itself and charge itself autonomously," says Prof Larson.
"At that point, wherever you are in the city, you can call for a vehicle. It can come to where you are and you drive off.
"When you reach your destination, you get out of the car, put it into autonomy mode and the car goes back to wherever it needs to be, ready for the next trip."
The idea is that such a system would make borrowing a car as convenient as owning one, but without any of the drawbacks, such as needing to find somewhere to park it.
New solutions, new problems
But the existence of driverless cars creates new problems.
Who would be legally responsible in the event of an accident, for example?
Even if the evidence suggests computers drive better, would people actually accept being at the mercy of a box of electronics?
And if cars are designed to communicate with one another, what is there to stop the system being targeted by cybercriminals?
Some answers are already being found.
In the Media Lab at MIT sits a model car known as Aevita. She has striking blue eyes - or rather, her headlights are surrounded by flashing blue optical fibres and they follow you as you move around the room.
She can also speak to you in a rather pleasant English accent.
The idea is simple. If future cars do not have drivers they are likely to make pedestrians nervous.
After all, how will you know whether it's safe to cross the road, if you can't look the driver in the eye?
Well, with Aevita, you can. Prof Larson thinks MIT's Media Lab has solved a problem that does not even exist yet.