Self-driving Cars and The Pingdemic
This edition takes a closer look at self-driving cars challenges, their potential implications for freight logistics and Australia’s supply chain shortages
If a self-driving car crashes into a tree, who’s at fault – the driver, the car, or the tree?
The discussion on self-driving cars has been reinvigorated recently following the Dawn Project full page ad “Don’t be a Tesla Crash Test Dummy”. Their argument is simple, there are risks associated with continuous vehicle connectivity and automating driving and until self-driving vehicle software is good enough, self-driving cars should not be allowed on roads. This discussion, although important for passenger cars, is critical for freight logistics as well. The truck driver shortages experienced in U.S., Canada, U.K. and much of continental Europe have prompted some to speculate that self-driving trucks will solve truck driver shortages. In a similar vein to the Dawn Project, I analyse some key considerations when considering self-driving cars and trucks: privacy, liability, and cybersecurity.
Privacy
Smartphones are as much surveillance devices as they are communication devices. Will self-driving cars be any different? While Tesla offers passengers the option to opt out from data collection from within the vehicle, cameras and sensors constantly scan and record the environment. Pedestrians outside vehicles do not have the same privilege to their own privacy. In effect, a fleet of self-driving cars can be used as a mobile CCTV system. Ideally these data should only be used for stated purposes – improving the self-driving algorithms. The temptation to use these data for surveillance may be too big for some companies and governments.
Liability
If a self-driving car crashes into a tree who’s at fault – the driver, the car, or the tree? Self-driving car manufacturers claim that even with the current technology levels, self-driving vehicles are much safer than human-operated ones. The key metric used by car manufacturers is accidents/ kilometre driven where self-driving vehicles are up to 3-times safer than human operated. Despite this improved safety record, self-driving cars do crash. Deciding who is at fault in a self-driving car crash is a challenging issue.
In the most recent cases involving self-driving vehicles, and following current legislation, drivers are held liable for incidents involving their cars. Increasingly, there is some reckoning over the fact that drivers cannot maintain high levels of situational awareness during long periods in which cars drive themselves. More importantly, the vehicle’s decisions are not visible to the driver in advance. So, even while maintaining full awareness a driver may not be able to react in time to a vehicle decision to break, accelerate or turn suddenly.
If it isn’t the driver’s fault when a self-driving car crashes, should it be the car manufacturer’s? The UK’s Law Commission supports this idea, highlighting in the Automated Vehicles Act that “manufacturers and software developers - rather than human drivers - should be held liable if a vehicle crashes when a self-driving system is operating”. Although, a welcome development, simply assigning the blame to the car manufacturer changes little. Who exactly is responsible - the chief executive officer or the software developer? To analyse the mountains of data coming from self-driving vehicles sensors, several car manufacturers use neural networks. Neural networks generally operate like ‘black-boxes’ – the input and output are visible but the process of transforming the input to output is opaque. There may not be many software developers that want to claim responsibility for the decisions of a black box.
Why was the tree there in the first place? In 2018, an Uber self-driving vehicle crashed into and fatally injured a pedestrian on the road. The investigation revealed that the "the [self-driving] system design did not include a consideration for jaywalking pedestrians". Because the pedestrian was not acting according to expectations (set out by legislation), it was practically invisible for the self-driving system. Unfortunately, these contextual differences (pedestrian on crosswalk, pedestrian not on crosswalk) are difficult to discover a-priori, mainly because of the inherent opacity of the software.
Cybersecurity
As the world becomes increasingly digital, cybersecurity breaches will continue to increase in frequency and severity. The Colonial Pipeline ransomware in April 2021 led to the operator suspending petroleum product deliveries until the ransom was paid. Toll group suffered a ransomware attack in 2020 while the company was in the midst of a 420 million dollar digital transformation project. Digitalisation can improve operational efficiency but can also expose businesses to cybersecurity threats. Because self-driving cars are continuously connected to a network, they too can become cyber attack targets. Imagine trying to open your car one morning and realising that it won’t open the door unless you pay 200 dollars. While most cyber attacks have targeted stationary computers, taking control of a self-driving car can transform it into a remote-controlled missile.
What do self-driving cars have to do with supply chains and logistics?
Truck driver shortages have been front and centre in recent supply chain disruptions and some news outlets have suggested that self-driving trucks may well be a solution to this problem. However, the privacy, liability and cybersecurity challenges outlined above with respect to cars, will play a key role for trucks. Self-driving trucks will likely record and transmit driver, passenger, and environment information. Self-driving trucks can cause much greater damage than a vehicle in case of a crash and, if their security is compromised, can become deadly remote-controlled projectiles.
Electric vehicles and self-driving capabilities often come as a package. Because electric vehicles are “good” for the environment, everything associated to electric vehicles must also be good. Self-driving capabilities do not necessarily have to come with electric vehicles, certainly not until these the privacy, liability and cybersecurity issues are sorted.
Australia’s Pingdemic
Return to supply chain normality, government says no
Australian supply chains are tested once more, this time due to rising COVID cases and self-isolation requirements. Empty supermarket shelves, closed stores are a common sight both in major cities and rural and remote regions. If all of this sounds somewhat familiar, it should. Just 6 months ago, the UK was facing its own pingdemic – workers being instructed to self-isolate and get tested because they were listed as potential close contacts to positive COVID cases – with exactly the same results as in Australia today.
Just like in the UK 6 months ago, Australia’s supply chain crisis is entirely created by the poorly designed COVID-related rules. Governments have tried generalised lockdowns which paralysed supply chains. Although lockdowns are now individualised (self-isolation rather than generalised isolation), their effect on supply chains is rather similar. What is surprising is that in an age where information is so easily accessible, foresight isn’t. I struggle to understand how governments fail to consider the potential outcomes of their decisions before making them – especially when such decisions have implications such as potential famine and shortages of medical supplies.
In other news
A few weeks ago, I re-posted a video of container trains in Los Angeles being raided by thieves. This piece explores option to combat rising theft from trains, from improved rail car design to loading containers door to door and rail companies employing their own police. If theft continues to mount, it may well be that rail companies might soon employ tail gun operators on trains. Someone on Reddit illustrated this concept on school busses.
Shipping containers are quite sturdy – they can be stacked 7 or even 9 high, some people use them to build houses or pubs. So, it may be a bit surprising to see them crumpled like cardboard boxes. This is what can happen during rough seas.
A really interesting design for a gas turbine truck by Ford in the 1960s.