What challenges do driverless cars actually solve?

Following on from the mobility panel discussion at tonight’s Vivid event, I thought its worthwhile evaluating the real world impact of driverless cars. There’s many touted benefits, often imagined...

Following on from the mobility panel discussion at tonight’s Vivid event, I thought its worthwhile evaluating the real world impact of driverless cars. There’s many touted benefits, often imagined as automatic outcomes, but in reality, such a dramatic change doesn’t happen without a lot of people working really hard on the problem. So what problems will driverless and autonomous vehicles actually solve ?

City congestion (yes)

Our cities are increasingly congested and instead of spending time working or with our families, millions of Australians are spending far too long sitting in traffic. One of the biggest causes for this congestion is the volume of cars, but the growth in population is likely to counteract any reduction in vehicles as we move to driverless cars. The theory goes that our dormant cars that sit in our garages at night or in the multi-story car parks while where at work, could be eliminated once we’re all rolling around in driverless cars. Cars would simply pick us up and drop us off, then go pickup the next person, then the next. In theory we could see a reduction of the overall number of cars, but congestion is caused by a lot of factors.

Driverless cars will always use route guidance technology, something auto makers and smartphone makers have been unable to achieve with humans in any meaningful way. Far too often drivers who could have avoided a heavy traffic area, drive straight into it because they’re relying on local knowledge, instead of leveraging connected datasets that would dynamically re-route in the event of an accident or congestion traffic area. This means less cars on the same pieces of road and that’s important to getting you to your destination faster.

Speaking of accidents, there’ll be less of them, and breakdowns. While autonomy can be implemented to vehicles with combustion engines (please don’t) most driverless vehicles will have an electric drivetrain. With pure EVs (not this hybrid rubbish) containing far less parts, there’s less opportunity for things to go wrong, which results in less breakdowns interrupting traffic flows.

Another big cause is driver’s slow to take off from a set of lights or intersections. At times its caused by human reaction times, which autonomous cars will easily beat humans at, hands down everyday of the week. The other cause is driver distraction, usually caused by checking your phone. This also gets completely solved when humans aren’t required to operate the vehicle (level 4 autonomy).

On the walk back to the hotel tonight I noticed another major problem in our cities. Bloody traffic lights. They are the dumbest implementation of traffic lights possible. Most intersections are on incredibly basic timers. These take no notice of the number of pedestrians waiting to cross, or the traffic flow at that time of the day. This means millions of people are being held longer than necessary and its an issue I hear nobody talking about fixing. The biggest innovation in traffic lights has been the introduction of count down timers until the change, but this is simply a physiological solution, to calm the driver, rather than a real solution. There are some intersections with sensors in the road that can turn a red light green, but due to the civil works required and the expense, these don’t scale. We need a vision sensor system that’s embedded into the lights that dynamically change the sequence based on the traffic flow. Tonight I witnessed a 4-way intersection with at least 50 people wishing to cross, with no cars for at least 1 minute. People waited, got sick of waiting, then crossed anyway, not because they’re criminals, but because this is technology failing at solving congestion.

Safety (yes)

There was a suggestion during the session tonight that we may need driverless zones in cities and unless your vehicle is being driven by computers, you can go in there. This won’t be necessary. The technology as shown by companies like Nvidia, Tesla, Google and others can understand the environment they’re in completely and respond accordingly, instantly. Make no mistake, there’s a race on right now, to build the best technology that results in a car that effectively can’t be crashed and can’t be crashed into. It’ll happen, the question is when and there’ll be a big first mover advantage (both in public opinion and in sales) to the company that can prove it works and ship product.

We’re headed for a pretty harsh transition period as cars will land in one of two camps, before and after level 4 autonomy. Once this technology becomes available and is affordable (another big challenge) good luck selling your non-autonomous car 2nd hand, it’ll be the one that can still have crashes. Along with the laundry list of other benefits autonomy brings, this will be the biggest hit and essentially kill the second hand car market.

With more than a million people globally being killed every year (1.3 million to be exact) by automotive related incidents, safety is a massive motivator to accelerate the release of vehicles that don’t require human control.

There was also a question about the morality of self-driving cars and that age old question of who should they kill, the person who paid money for the car, or the people that didn’t? This is obviously a false question that is answered by the technology being right. Vehicles will understand their environments completely, in 360 degrees and with multiple levels of redundancy in the instruction set, monitors of vehicle hardware capabilities, the car will also stop without injuring either the occupants or pedestrians.

Mobility (maybe)

Australia is like many other countries which have an aging population. That means the number of people in their older years is growing faster than people in their younger years which creates a unique transport challenge. This is an easy one. Not requiring a human to drive a vehicle opens the door for the elderly and disabled who can’t drive due to physical limitations, still maintain independence and visit family, get to the shops or doctor. This is an awesome side-benefit of the technology, but often the elderly don’t have a lot of cash, so increasing the affordability of the technology is critical in impacting this segment of the market.

Its here where companies need to innovate on their business models to get expensive vehicles in the hands of those who can least afford it. Its likely many businesses will be the first to purchase autonomous vehicles as they can distribute the cost over a number of customers. Imagine a nursing home that currently buys 5 specially kitted out mini-van with a wheelchair lift and of course employs a driver. Instead that same nursing home may decide to invest in 2 driverless mini-vans and be able to service the same volume of customers. Until there’s actually vehicles shipping into Australia that have hard prices, its hard to say the business model stacks up, but its at least plausible.

Cost of ownership (maybe)

The biggest indicator we have so far that autonomous vehicles that you buy personally won’t be able to be operated for profit is Tesla’s terms and conditions. Hopefully other car companies will try and gain marketshare by differentiating on this policy. For those able to afford the expensive early models, its feasible you can make a mini autonomous taxi-company, making you money when your car’s not in use, as it takes other people for a ride and sends you some $$ for the use. Only problem is you’ll likely be competing with the likes of Uber who has already publicly stated they’ll replace humans with driverless cars when possible. This takes a massive cost from their business out of the equation, so there’s a massive incentive to make it work. This also opens the door for Uber’s to be everywhere, not just in cities where current economics and ratios of drivers to riders make sense.

The cost of owning a car right now is pretty high when you consider registration, which by the way Governments should start planning in a fall in revenue from this, then add insurance, maintenance costs, fuel, cleaning and more. The overall cost on a family budget could go down if the traditional 2 car model makes room for a single car that takes the family where they need to go, but that relies on no 2 family members needing to be at different destinations at the same time. Its easy to see some families deciding they’ll pay a few dollars every time they need to travel, rather than own a car at all. This will become an easy calculation to make as we get better at tracking financial expenses and understand average trip costs using autonomous vehicles. As of right now, with the costs of fully driverless cars expected to be high, its possible the numbers don’t yet stack up for a saving column of your annual home budget.

Eliminate infrastructure upgrades (No)

With cars using existing roadways more efficiently, its possible Governments will be able to delay the upgrade to road infrastructure. That’s the thing, its likely to only delay an upgrade, not avoid it. Assuming Elon Musk’s network of underground tunnels don’t happen in Australia anytime soon, Aussies may also decide that moving around using driverless cars is so much of a better experience that we do it more often.

If we all continue to order more products online and those deliveries are made using vehicles, services like Deliveroo will potentially increase the overall number of vehicles on the road. Pair this with continued population growth and the answer to this one is absolutely not. The estimates are that by 2061, Australia will have a population of 40 million and 9 out of every 10 Australians will live in East Coast major cities. Personally I’m a big advocate for working at home, or in regional cities, but neither seem to be happening in any meaningful way.

Jobs (no)

The development of autonomous vehicles takes tens of thousands of really talented engineers, computer programmers, data scientists and other specialised roles. None of that makes up for the sheer number of taxi drivers, truck drivers, delivery personnel that will be relieved of their duties once vehicles drive themselves.

If you’re thinking these people will be fine if they can just buy their own driverless car and send it to work on their behalf, the reality is, 1-1 ownership won’t cut it. These cars will be expensive to start with and that’s why companies like Ford are targeting large companies, rather than individuals with their first releases due in 2021.

As you can see from the items above a large part of the autonomous revolution relies on the cost of the vehicles capable of it, being affordable. If anything its not the technology that’ll be the differentiation, it’ll be the accountants. There’ll be big dividends for the company who can ship first (must be safe) in public opinion and potential fleet sales from mid-to-large enterprises and governments.

If there are other benefits or problems that I’ve forgotten, leave your thoughts in the comments.


Disclaimer: Jason attended the Vivid session as a guest of Ford Australia.