Not Such A Long Road To Driverless, Claims Kar-Go

Many automaker experts believe fully autonomously driven vehicles as a feature of every day life on our roads is much further away than that being assumed by their digital tech peers.

Head of Toyota Research Institute, Gill Pratt said as much at CES 2018. “It’s a mistake to say that the finishing line is coming up very soon. Things are changing rapidly but this will be a long journey.” This long journey could take us beyond 2050 according to what many auto insiders have told me.

Yet William Sachiti, founder of autonomous last mile specialist Kar-Go is firmly siding with the optimistic techies believing advances in computing capacity will bring that future much closer. Speaking exclusively to TU-Automotive, Sachiti said: “What happened was the GPU [graphics processing unit], this came out of nowhere. If you ask people in the technology space, most of whom were at university around 2005, they did not pick up on artificial intelligence for a long time because they knew that we didn’t have the computing power. However, when the GPU was invented, something that can do so many calculations so much quicker and, today, it enables us to do over seven trillion calculations per second.

“This is the reason autonomous cars are beginning, finally, to work as they should. If you look at the people really leading the charge to autonomous, they are not automakers because it is not a car but a computer on wheels. This is computer science rather than auto engineering. We are a tech company and we are involved with autonomous cars because they are tech. This is why the car companies are trying to play catch-up by trying to hire tech people because it’s a tech-play not a car-play.”

Sachiti said no one corporation can steal much of a march on the others because the autonomous technology is built on science that is shared as soon as it is discovered. “For example, we use a specific type of super computer to do all the neural processing on a driverless car – this is the very latest in super computing for driverless cars and no-one on earth has anything more advanced because the science is not even possible.”

However, he admitted that the safety issue around testing current driverless technology on open roads can cause concerns but that the end goal is worthwhile. He explained: “Getting autonomous cars on the road, will it be safe? It will never be 100% safe but it just needs to be better than humans. So, if we have 1.2M auto deaths a year from humans driving and we reduce that number by half with autonomous cars that’s a win. With the tests we do, we are pretty much on par with humans in an uncertain world and what I mean by uncertain world is not simulation but on the actual road. When we get to being four to five times as good as a human driver, I think people will start to accept it more.”

From a business perspective, the opportunities that autonomous technology affords are too attractive to ignore, said Sachiti. He explained: “The second way of making money follows a report by McKinsey earlier this year saying that the autonomous car space will have a value of $1.5Trn by the year 2030 – one of the most valuable industries on earth just in the of data only. So for us, data is our biggest pledge, the amount of data we have and what it means is so valuable.

“Just for what we do, autonomous cars for last mile delivery, there will probably be two or three businesses born out of what we do. An example would be if you remember milk daily deliveries but the rising costs of delivery made that harder to do as a business. When we make autonomous deliveries costing less than £1 ($1.30) for deliveries all day, where you can have milk, bread, razor blades, anything delivered daily, these are new types of businesses that will piggy-back off transport being so cheap. Data will drive that.”

— Paul Myles is a seasoned automotive journalist based in London. Follow him on Twitter @Paulmyles_


One comment

  1. Michael Haines 19th November 2018 @ 8:07 pm

    Ubiquitous delivery requires that most freight travel at night and our roads become managed so that at the limit any car wanting to enter the network must wait until there is capacity – with priority for emergency services and critical supplies. This is a very long way off. Otherwise every extra vehicle on the network just slows it down. The sort of image projected in this piece would put any large city into gridlock

Leave a comment

Your email address will not be published. Required fields are marked *