Part 1: What is an Autonomous Car And How Far away is it.

 

My son turns 10 this year. I have already told him he may never have to learn to drive. No balancing the clutch; no ‘Tennis’ neck pain from those every-5-second-blindspot checks; no parallel parking between those wonky white poles (not everyone can); no buying the K53 learners’ tests in the parking lot behind the Licensing department.

 

By the time he turns 17, the age when I got my learners license, self-driving cars could well be the norm. Going from point A to B would merely require you to step into the vehicle, enter your destination on the navigation system or your linked smartphone, and sit back and enjoy the ride. I have also told him that he may never know the joys of an internal combustion engine, be it petrol or diesel. The world is steadily moving over to electric vehicles, and certain progressive countries have already announced a ban on internal combustion engines from as early as 2030.

 

Whether we like it or not, rapid changes are coming to the automotive industry. Driverless technology is coming, it just being a matter of when. For driverless technology to become mainstream, much still needs to change.

 

In this article, let us take stock of where we are with driverless technology, and what hurdles still need to be overcome before  driverless cars - although already in testing in many parts of the world - become available to the general public.

What is an Autonomous Car?

An autonomous car, by definition, is a vehicle that is capable of sensing it’s environment and operating without human involvement. A human passenger is not required to take any sort of control of the vehicle at any time, in fact a human passenger is not required to be present at all. An autonomous car can go anywhere a traditional car goes and do everything that an experienced human driver does. (1)

 

There are currently 6 levels of driving automation defined by the Society of Automotive Engineers (SAE), ranging from Level 0 (fully manual) to Level 5 (fully autonomous).

Asking for a friend, what is the difference between Self-Driving, Automated and Autonomous?

 

A Self-Driving car can drive itself in some or even all situations, but a capable human must always be present and ready to assume control.  In the graph above, self-driving cars would fall under Level 3 (conditional driving automation) or Level 4 (high driving automation).

 

Think about setting the cruise control on your vehicle – your car is now driving at the speed you desire – but you are still very much required to override the cruise control when the situation calls for it, such as someone braking suddenly in front of you or an errant animal running across the road.

 

 More modern cars with Advanced Driver Assistance Systems (ADAS) may be able to brake for you too, but you still need to be present to steer the vehicle. Advances in self-driving cars mean they may be able to steer for you too, in ideal conditions, but cannot well-account for human stupidity and uncanny accidents.

 

In the example above, you can be driving down the N1 with your cruise control set to 135 Km/h (True speed 127 km/h, not legal but not fineable either, thank me later). Your vehicle has automatic emergency braking and for arguments sake it can steer itself too, when the sprightly caravan ahead of you in the alongside lane de-hitches and comes spinning towards you! You still need a driver to quickly intervene and make the split-second decision of what to do avoid an accident.

 

Automated and Autonomous are both superior in intelligence to self-driving, the difference being the degree of human intervention. An automated car does not have the level of intelligence or independence that an autonomous car has (2).

 

The word autonomy has implications beyond the electromechanical. A fully autonomous car would be self-aware and capable of making its own choices. For example, you say “drive me to work” but the autonomous car decides that you are overworked and underpaid and takes you to the beach instead (Give that software engineer a Bells). A fully automated car, however, would follow orders and then drive itself to your office. (1) On some levels, an automated car would be a Labrador and  an autonomous car a Terrier.

 

A Self Driving car is also subject to geofencing, unlike a fully autonomous Level 5 car that could go anywhere.

 

 

How Do Autonomous Cars Work?

Autonomous cars rely on a complex ecosystem of sensors, actuators, complex algorithms, machine learning systems, and powerful processors to execute software.

 

In essence, autonomous cars build a real-time map of their surroundings based on a bevy of sensors situated in different parts of the vehicle. Radar sensors are used monitor the position of nearby vehicles and certain fixed objects. Video cameras are employed to identify traffic lights, read road signs, track other vehicles, and watch for pedestrians. Lidar (light detection and ranging) sensors bounce pulses of light off the car’s surroundings to measure distances, detect road edges, and identify lane markings. Ultrasonic sensors, housed in the wheels, distinguish curbs and other vehicles whilst parking. (1)

 

Sophisticated software then processes all these data-feeds, plots a path, and relays instructions to the car’s actuators, which control the vehicle’s throttle (acceleration), braking, and steering systems. Hard-coded rules, obstacle avoidance algorithms, predictive modelling, and object recognition help the software follow traffic rules and navigate obstacles. (1)

 

 

Challenges facing autonomous cars?

Fully autonomous (Level 5) cars already exist and are being tested and trialled in numerous parts of the world. But we are still a few years away from being able to purchase one on a showroom floor nearby. The challenges range from the technological and legislative to the environmental and philosophical. Let us look at a couple of the unknowns.

Lidar and Radar

Lidar is costly, and that’s Costly with a capital C. As a technology it is still in it’s infancy, and has yet to strike the right balance between range and resolution. There are still many unknowns. What if multiple autonomous cars were to drive on the same road, would their lidar signals interfere with one another? How many radio frequencies are available? If each vehicle used a different frequency, will the frequency range be enough to support mass production of autonomous cars?

 

 

Weather Conditions

In an ideal world, skies are blue, roads are black and lane markings are white or yellow.  What happens when Giloolys is flooded and under 15cm of water? How will the cameras and sensors track lane markings if the markings are obscured by water, oil, ice, or debris? How will the video cameras work in the early-morning zero-visibility conditions between Witbank and Middelburg? What will an autonomous car do if there’s a layer of snow on the road and lane dividers disappear? What if you are in the Eastern Cape and there are no lane dividers, as the contractor absconded with R27-million pre-payment and the SIU are investigating?

 

 

Traffic Conditions

We also don’t yet know how autonomous cars will handle bridges or tunnels. In early tests, cars simply refused to go into tunnels, seeing the tunnel walls much the same as concrete barriers or bridge supports, which, according to the software code V1.1.3, are best avoided.

 

Imagine travelling 90-percent of the way to Cape Town and your car simply refuses to enter the Huguenot Tunnel, instead deciding to drive back to Beaufort West, down to George and then a further 5 hours along the N2 to avoid any other such tunnel thingies. By the time you reach Cape Town you will be ready to push your car off Lions Head and claim from insurance, but it won’t even let you do that, because it’s Lidar controlled automatic handbrake detects your murderous thoughts and won’t release.

 

 

Laws and Regulations

This may not be a huge problem in South Africa, where the laws and regulations are very flexible (read non-existent, non-enforced and or generally ignored). But in the US for example, different States have adopted different laws regarding the use of autonomous vehicles. This makes interstate travel a headache. You can expect the DA to do something similar in Western Cape which again means a fully autonomous car for your JHB-CPT December holiday is a poor idea.

 

Some of the more colourful laws and proposed legislation coming out of states in the US include: A per-mile tax on autonomous vehicles to prevent the rise of “zombie cars” driving around without passengers; a bill proposing that all autonomous cars must be zero-emission vehicles; and legislation proposing that automated cars have a panic button installed. It’s unclear to me what the panic button will achieve in a car which doesn’t have a steering wheel. (Perhaps the senator proposing it was Republican, Texas)

 

 

Accident Liability

Who is liable for accidents caused by an autonomous car? The manufacturer? The human passenger? What if the passenger is not even human – but the family Labrador on a pre-programmed trip to the Vet? How can you throw the book at a harmless, cuddly, wet-nosed Labrador? And, in mitigation, the poor dog was sick at the time? 

 

The latest blueprints suggest that a fully autonomous Level 5 car will not have a dashboard or a steering wheel, so a human passenger would not even have the option to take control of the vehicle should an emergency arise.

 

 

Artificial vs. Emotional Intelligence

Human drivers rely on subtle cues and non-verbal communication— such as making eye contact with pedestrians or reading the facial expressions and body language of other drivers— to make split-second judgment calls and predict behaviours. We also know and identify hazards based on experience. That’s Mrs Smith, she can’t drive! Why, if that’s not Mr Cohen who wrote off his son’s Maserati? And of course, we begrudgingly submit ourselves to the King of the road, the Toyota Quantum.

 

Will autonomous cars be able to replicate this connection? Will they have the same life-saving instincts as human drivers? Will there be an algorithm which detects Toyota Quantams (and its Foton, Fudi, Jinbei, Chery, CMC and BAIC clones) and takes into account that they are highly likely to skip red robots, drive against oncoming traffic, stop suddenly in front of you and cut you off going straight from the turning only lane?

I would definitely add a few lines of code to account for these exciting possibilites, where the meanings of traffic light colour, road markings and direction of traffic are not fixed but random variables. What about when you driving in downtown JHB and surrounded by hundreds of Quantums? The possibilities could be endless.

 

 

Ethical Decisions

Automated Vehicles will have to be able make judgments with potentially life and death consequences. In the event of an inevitable accident, whose life should be saved and whose life should be sacrificed?

 

It is a modern extension of the well-known ethical dilemma or thought experiment known as the ‘trolley problem’. In the case of an automated vehicle, consider the following scenario. One is in an automated vehicle, which is driving down the road, in KwaZulu Natal obviously, when the vehicle notices the road has been washed away by the recent floods. There isn’t enough time or space to stop the vehicle.

Cameras, Radar and Lidar work overtime and come up with the following options:

1) Continue straight and the vehicle plus passengers go over the embankment and down the deep ravine, possibly fatally.

2) Swerve left into the sugar cane field saving the occupants of the AV, but mowing down the lady selling Avo’s and Mangos (and Litchis in season) alongside the road. She is also a single mother supporting 2 orphans.  

3) Swerve right where SANRAL workers are standing (to estimate repairs of the said sinkhole and/or buy Avo’s). At least two workers are in harms way.  Who gets to live? Who get’s sacrificed?

 

This ethical side is still stuck in dicey territory. And this is because humans and AVs make ethical judgements that are fundamentally different. The crash avoidance behaviours are hardwired in humans, and our reflexes kick in within 2 seconds in the face of danger. But can humans be held morally responsible for their involuntary reactions to such situations?

 

On the other hand, automated vehicles are outfitted with sophisticated sensors and algorithms to predict and react to collisions better than human drivers. However, AV judgments don’t take the ethical repercussions into account, not that there currently exists any consensus on the AV ethics code as of yet, and their decisions are pre-programmed, which is basically pre-meditated. (3)

In Summary

Autonomous vehicles (AVs) are one of the most polarising technologies primed for mass adoption. They have huge potential in terms of improving mobility, lowering driving costs, emitting less pollution and creating more relaxation time for drivers. By all accounts,  AVs are on their way to all good dealerships near you.

 

Just how far away? Read part 2 in next week’s newsletter.

 

References

  1. https://www.synopsys.com/automotive/what-is-autonomous-car
  2. https://www.technologyreview.com/2022/05/27/1052826/ai-reinforcement-learning-self-driving-cars-autonomous-vehicles-wayve-waabi-cruise/
  3. https://analyticsindiamag.com/the-moral-dilemma-of-embedding-ethics-into-autonomous-vehicles