A Wired story talks about the Otto retrofit solution where existing trucks can become autonomous vehicles for only $30,000.
I believe it’s a work in progress, but an aftermarket upgrade that can convert an ordinary truck (manufactured after 2013) into a driverless truck for $30,000 doesn’t sound like a bad investment.
From what I can glean at this stage all the prototype does is stay within a lane. It can’t change lanes without a driver, but will slow down and platoon behind anything ahead in it’s lane. A bit of a shame if that’s a vehicle traveling at very slow speeds.
However this is a project in development with some high flying leadership from companies like Google Maps and Tesla.
It’s sort of comparable to a plane that requires (legally at least) a pilot to take off and land. This truck requires a driver to change lanes and I to manoevre through intersections and complex road networks. It could however give a driver a lot more rest time when they are on the freeway which would have them being more alert when needed.
What I feel a lot of previous news stories have been a bit misleading is countered in the following statement from the article: “Google’s self-driving cars, for example, can’t go anywhere without extremely high-res maps in hand”. These don’t exist in many parts of the world, or at least don’t exist commercially.
Having driven guided by car navigation systems from leading brands in New Zealand, Australia, the USA and Europe. I have had many experiences where I’ve been on roads that aren’t on maps, where the maps are inaccurate and where there have been significant changes to roads such as realignments. This despite navigation brands claiming to have highly accurate maps that are frequently updated.
If you look through my blogs you will find countless stories of accidents and mishaps where humans (who are in control of vehicles) have driven into rivers and canals, got stuck between buildings on narrow roads and more; and that is with a driver behind the wheel. Here are just 18 examples http://bit.ly/25bn9DL. If you troll through my blogs you will find many more. #TheGPSMadeMeDoIt
A human should be able to recognize that what they are being told is different to what they are seeing. LiDAR and other technologies ate pretty good, but they do not work perfectly and not in all conditions which is why mapping cars use a combination of several technologies. HD Mapping is still relatively new and in many countries limited to small parts of the roading network.
The concept is fantastic and there are distinct benefits from low cost retrofit systems, however IMHO it is going to take several years before we see driverless vehicles becoming a trend in any significant numbers (outside of specified freeways) and if the technology is pushed to hard and fast to be commercialized, the industry could be significantly set back due to fall out after accidents or crashes. This is why truck manufacturers are involving insurance companies as supporters at an early stage.
It’s fine to use insurance companies as a reference at this point and good, well meaning publicity, however one must remember that insurance companies are risk averse and will change their mind in an instant if they find they are paying out on significant claims.
Great article Luigi. I think the bottleneck in the end won’t be technology advances but risk management and legislation which poses the question “who exactly is to blame when an accident occurs?” Because accidents will occur at some stage. The thought experiment of the trolley problem brings this to light. If the computer detects an impending accident and calculates that the car will kill several people in another car or swerve and kill the driver, what will it do???
LikeLiked by 1 person
Thanks Daron, you’re right accidents will happen. I think we’re a long way from AI thinking that far which is a bit like a security AI in a terrorist environment seeing someone wearing clothing that hides their faces and decides they are a threat. I think the biggest problem is ignorance and executives jumping on a bandwagon they know very little about. When I read stories from Volvo saying don’t worry, insurance companies will totally cover anyone who operates one of their driverless vehicles. Sure they will, until the first serious crash where there were no other parties involved. What happens to everyone who has invested in this technology when that happens. I have cancer as you probably know. The first thing my insurance broker said to me was, Luigi, you are now no longer insurable, whatever you have is the most you will ever have. Effectively the circumstances changed. So will Volvo’s underwriters cover a truck that crashes on Haast Pass where there is no mobile signal and the road centreline in patchy? Will they cover a vehicle that drives off the road when it can’t see the centreline or the side of the road because of ambient sunlight and a wet road? Will it cover the owner when speed data is incorrect and the vehicle drives at 100kmph in a 70kmph zone and gets ticketed or crashes into people or vehicles that would have been perfectly safe if it had been traveling at legal speeds?
Pingback: Remember the jokes about what if computer manufacturers built cars? | SoLoMo Consulting