Killer cars? Or the end of human error?

How close are we to a driverless future?


Self _driving _Uber _prototype _in _San _Francisco
Next Post

The tragic death of a pedestrian killed by an autonomous vehicle being tested in Pheonix Arizona will, no doubt, be investigated thoroughly. Whilst awful accidents do happen, most human drivers would hit the brakes or swerve on seeing someone in front of them - and then the consequences are less likely to be as bad. After all, road safety in the UK has shown that the difference between being hit at 30 and 35mph can be the difference between life and death. The questions around this sad accident boil down to ‘did the vehicle fail to detect the person crossing in front of it?’ or ‘did it detect the person but fail to brake?’. 

Whilst forensic investigations take place, we decided to look at how, whether and when - autonomous vehicles will hit our streets in the UK. 

Well, the first news is that they are already here, although in this case the ‘streets’ are actually pavements and they’re in Milton Keynes. 

LUTZ Pathfinder Autonomous Vehicle UK

Image: Pathfinder in Milton Keynes (Transport Systems Catapult)

LUTZ Pathfinder vehicles were developed by RDM (an automotive engineering company in Coventry) and the Oxford University Mobile Robotics Group and are being live tested on pedestrian areas of Milton Keynes. They’re capped at 15mph and are supposed to move at ‘no more than a brisk walking pace’. Why on earth we’re testing vehicles which really aren’t very useful unless they’re on the roads on pavements is anyone’s guess. But in the meantime these two-seater pods have been testing their cameras, LIDAR (laser scanners) and radar based detectors’ ability to navigate and not run people over for over a year now. Initially with a human operator on board, they have been spotted running in fully autonomous mode. The plan is that these vehicles will form the basis of a fleet of autonomous taxi pods.

Waymo Driverless Car

Image: Waymo driverless car, Grendelkhan

When it comes to autonomous vehicles on the actual streets, they seem a little further off. Whilst some in the industry predict that it will be 15 years before we see fully autonomous vehicles on road, our Secretary of State for Transport anticipates seeing ‘fully self-driving cars’ on our streets by 2021.

To this end, the government has introduced legislation covering autonomous vehicles - the Automated and Electric Vehicles Bill which is currently before the House of Lords (March 2018). This requires the government to keep a list of all the makes and models which are licensed to drive on UK roads in fully autonomous mode. It also extended the insurance framework from insuring drivers to include insuring autonomous vehicles (with a number of provisos to cover the insurance frameworks if vehicles are hacked - and indeed to allow insurers to sue vehicle manufacturers if a flaw in the vehicle caused it to be at fault). 

This legislation is scheduled finish its passage through parliament over the course of this year - and will likely become law within 12 months. It covers fully self-driving vehicles - level 5 on the industry scale of autonomy between ‘no help to human driver’ (level 0) and ‘no human intervention at all’ (level 5). 

With the various estimates of when we’ll see level 5 vehicles on the streets, it’s a useful exercise to look at what we’re driving now (the Milton Keynes Pathfinders apart).

- Cars have been able to parallel park themselves for over 15 years. The first model for sale was the 2003 Toyota Prius although trial vehicles were developed before this.  

- Lane monitoring was introduced in 2001 and actually assisting vehicles to keep centred on motorway lanes has been offered since 2006 (the first system was introduced by Lexus). 

- If you go to a new car launch these days, there will inevitably be a moment when all the petrol heads glaze over whilst the presentation raves about their latest offering’s ability to communicate with other cars and be warned about crashed or stopped vehicles ahead so they don’t run into them. Connected vehicles are a further step to autonomy. Whilst humans are limited to the information they can see and hear, connected cars can share information with each other - enabling them to know what’s round the corner and avoid congestion and accidents. Connected vehicles should also be able to ‘platoon’ - form long trains of vehicles which can be driven very closely together because their operating systems are communicating so they brake and accelerate in unison.

- Tesla introduced it’s ‘autopilot’ mode in 2014. Lane assist, automatic lane changing and maintaining distance between vehicles enable cars to cruise down motorways with startlingly little supervision - if you don’t leave your hands on the steering wheel though, they tend to get snarky with you. This is because the legal framework requires a human is there to intervene in case of emergency (not to mention that they’re not 100% accurate in abnormal conditions).

Whilst these are all working technologies which encourage us to see full autonomy may not be far off, they’re generally quite limited to controlled driving conditions. The fully autonomous vehicle tests Uber was conducting in Pheonix Arizona were on dry, wide open US roads - not the kinds of driving we see in the UK. However, there are multiple other players in the field - from auto manufacturers like GM to Tesla as well as the likes of Google and Apple - some of whom are testing in California as well.

Will we be seeing fully autonomous vehicles on our roads by 2021 as the Transport Secretary suggests? The jury is out - but it’s clear that when they do arrive they’ll need to be able to deal with the wide variety of roads and driving conditions we encounter in the UK.

 

Main image: Self driving Uber, San Francisco By Dllu (Own work) [CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0)], via Wikimedia Commons


By Beate Kubitz at 23 Mar 2018, 00:00 AM