Self Driving Cars – Are They The Future?

The automobile has been with us for more than 125 years and has undergone thousands of iterations, but one factor has remained the same – the Driver. With the technology, infrastructure and skills at our disposal, can today’s society completely redesign the concept of the car to be self-operating and self-navigating? Many argue that the last obstacle in making the car seamlessly ubiquitous in our lives is removing the human factor. There are not many other investments which could potentially save hundreds of thousands of lives per year, increase worldwide productivity, save scarce energy, reduce traffic jams, improve physical space, reduce the cost of transportation and generally make our lives better, that are as accessible as automating the act of driving.

As a car enthusiast who thoroughly enjoys driving in it’s various guises, both on and off the racetrack, I understand the unadulterated joy and freedom that comes from piloting a car to one’s potential. In my mind, there will always be a place for human-controlled driving (previously known just as ‘driving’), but unfortunately one that increasingly will be relegated to leisure or niche pastimes, rather than a necessity. Even as I ache for the cars of yesteryear, it’s an activity which I continue to enjoy until it becomes prohibitively expensive to do so. For the purposes of this article, let’s discount myself and the few percent of enthusiasts within the population. The truth is, a self-driving car will be the biggest revolution in the automobile’s use since it’s inception and it has a very real potential to make transportation better, but it will require a monumental change in attitudes and beliefs – the car is such an ingrained symbol of our society, a way for so many generations to identify some part of themselves.

What we should consider is: “Is there a place in our future for self-driving cars as transport, and what does it mean for us?

Where are we at now?

For the more attuned, you may have noticed over the past decade or so, automotive development has led to more control being taken away from the driver’s hands – in the interests of safety and sales. Safety, because the plan facts are, the vast majority (more than 90%) of car accidents are caused by human error or misjudgement of conditions (more on this later). Mechanical failures or unforeseen circumstances are few and far between. Sales, because more accessible/convenient features and safer cars sell. Most people will never use the nine airbags on a Toyota Yaris, but they’ll sleep better at night knowing they are there.

Cars nowadays boast a full plethora of automated systems: Self-parking, lane sensing, radar-guided cruise control, blind spot detection, clever 4WD systems, dynamic stability/traction controls, not to mention constantly improving chassis, suspension and brake designs. In short, cars are becoming far easier to use, and much harder to crash and hurt occupants. It’s also important to note that there are an increasing amount of convenience electronics within cars these days: features such as social network and music player integration, not to mention wi-fi hotspots – there are even mobile apps on mobile platforms that control car functions and security. The trend is to make cars less about the driving, more about your lifestyle and complementing it with an always-connected vehicle. Read about systems from Toyota, Mini, Chevrolet, Cadillac, Nissan and BMW.

Naturally, to run these systems, greater and greater computing power is needed – today’s cars have hundreds, if not thousands, of electrical modules and sub-systems, that run every minute detail of the car’s functions. Advances in computing power and miniaturisation help to fit even more powerful systems and sensors into smaller spaces. It’s just a matter of time before cars will take over the act of driving completely. There are challenges, but there are also rewards.

What are the benefits of an automated system?

Much has been made about advances in self-driving technology, but sometimes it may be unclear what the goal is. In a utopian future, cars will function as a utility or appliance ( for the majority of people commuting / transporting). Just like the majority of people today would rather their smartphone is slick, easy-to-use and enables them to achieve greater functions – the majority of people want cars which fulfil the same needs. Once again, I’m in the minority where I would prefer to tinker to improve, but the case is clear – we need autonomous cars. So let’s have a look at the benefits of this ideal scenario, where an increasing number (or all) of cars are automated.

Safety – The biggest one is, of course, safety. Car manufacturers, scientists, statisticians and governments have long known that humans piloting cars around themselves is a recipe for disaster. Humans are fallible by nature, they have a limited information processing ability, a short attention span (cue texting / talking on phone), poor memory, fatigue, can often lack fine-motor skills (such as skid correction, or even as simply as pressing the wrong pedal), age-related failings, can’t see the in dark, have tunnel vision, form bad habits and are prone to emotional (think road rage) or inhibited driving. We’re good for lots of things, but racking up many incident-free kilometres behind the wheel is not one of them.

When you place the responsibility for transporting people within a mass of steel and aluminium travelling at high speeds (even the low-end cars these days can push 150km/h+), into the hands of such a fault-ridden meat bag known as humans, no matter how skilled they think they are, then bad things are bound to happen. 1.2 million people die each year around the world from car-related fatalities, the majority of which are driver-induced. Take control away from the main source of the faults, and the fatalities reduce (by an estimated 50%). Automated cars could save up to 600,000 lives per year, this is no laughing matter.

In contrast to humans, computer systems monitor the environment and car’s state hundreds of times per second, all day, every day and react in an instant to any adverse conditions. They can sense upcoming hazards, emergency vehicles and react in a thousandth of the time of a human. Think about the number of safety features in cars that have been implemented purely to correct human’s erratic behaviour or ham-fistedness.

In some cities, cars can’t be avoided – public transport is woeful or there is a very sparse population density – individual transport is the only viable option. In others, obtaining a driver’s license is comically easy, with drivers not being trained how to handle anything outside the bare necessities. The answer so far, has been to implement features such as stability control systems in cars or more speed cameras, though observations indicate that the greater the capability of cars, leads people to push the limit more readily – presumably satisfied that the computer will take care of the rest. In others, the car just holds too much cultural importance to be easily sacrificed.

Time – Commuting, especially in peak hour traffic, can take a huge amount of time for the average person. A standard 45-minute drive to work equals a full 360 hours per year, doing nothing but sitting in your car driving to and from (imagine what you could get done if you had an extra half a month a year). In the fast-paced, multi-tasking world we live in, time is at a premium. I know that, at least for commutes, I’d much rather be working/reading/sleeping/writing/doing something else than staring at the car’s bumper in front of me. Zen-like reflection rarely occurs in the red haze of peak hour.

Traffic – I’ve always believed traffic is about flow, not order. When you block somebody off from merging, or drive spontaneously without paying heed to the flow of traffic or traffic lights in the upcoming kilometre or two of your vision, you might be satisfying your ego, but you are adding to the traffic problem. These days, there are multitudes of traffic monitoring sources, providing instantaneous feedback on where traffic hot spots are occurring – this information can be seamless implemented into a car’s routing preference for optimum results.

A traffic engineer I spoke to indicated that a large amount of his time is spent with sophisticated computer models and micro-simulations to improve traffic flow across the entire road network. Each minuscule change in traffic light duration, lane markings and signage can have a huge flow-on effect through the network. A large barrier or accurate traffic simulation is the random behaviour which is attributed to each driver’s personality – each driver is interested only in their own advancement in the queue, rather than the benefit of the whole.

By having a completely automated system where all cars can communicate with each other and act to smoothly regulate traffic (well outside the capability of humans), then traffic in itself plummets and everybody benefits. Reduced requirement for traffic gaps, smooth and even traffic flow, even vastly increased speed limits due to all the above factors. Needless to say, road rage will become a rarity – people will be getting to their destinations quicker than before, at arrival times precisely estimated by the system and there’s nobody to blame for hold ups (if there were any).

Fuel economy – A natural off-shoot of the reduction in traffic and natural avoidance of built-up blockages is an increase in fuel economy. With less stop-start and better choice of roads (even today, GPS makers like Garmin can provide the choice of a more fuel-efficient routes based on hill gradients and estimated traffic loads based on historical data), fuel use drops, ‘nearly’ everybody wins.

The truth is, the average driver does not need a RWD, 200kw family car to commute. That extra power is there for two reasons: 1) In the rare situation where sudden unexpected overtaking is required [mostly due to poor foresight] or 2) On the brochure, to sell more cars. This is where the future of electric cars ties in – the instantaneous torque of electric cars lends itself very well to a large network of automated nodes. Energy regeneration and expenditure can be exactly calculated on a global scale and vehicle behaviour altered to achieve maximum energy efficiency.

Parking / Convenience – The bane of the modern driver is finding a parking spot, whether it be in crowded city streets or at the suburban shopping mall. An automated car will be able to drop you at the doorstep of your destination, then travel off and find a suitable safe parking spot. Contact your car later (assuming it has network reception) and it meets you at your current or pre-defined location and you’re off. Parking door dents, parking spot rage and endless circling are all eliminated in one fell swoop. Imagine how many more cars you could fit into a parking complex, if you had the ability to shuffle cars out of the way at any time. Even the too-young, the too-old and the disabled will have readily available convenient transport.

Taken a step further, if cars were an indifferentiable utility, then the entire network of cars would be at each person’s disposable. Effectively, you’d hire a nearby car to take you to your destination, then pay your hire fee and carry on – an extensive trouble-free taxi system at your fingertips. If your car breaks down, jump into another one. In a foreign country? Transport is waiting. The city-owned (or private group-owned) cars would save individuals from the costs of depreciation, maintenance and insurance, whilst allowing for far more efficient usage of space and resources in dense areas like cities. Many people don’t even use their cars very much, having all these resources and physical space tied up for an on-demand feature can be replaced entirely with a service.

Unfortunately, this type of system would require a complete change in culture. Our undying affinity for ownership of shiny objects in this consumerist age must change. The developed world has an addiction for brands which cannot easily be shaken. Ideally, just think Zipcar, but self-driving. With these vehicles so heavily loaded with tracking technology, and generally near maximum use all the time, car theft will be a thing of the past.

Space efficiency – Look around inside the modern car and unsurprisingly, you’ll notice how much space is dedicated to equipment that helps you drive the car. The steering wheel, pedal box, dash cluster, gear shifter, mirrors for example – with the remainder of functions tightly knitted around the driver’s position. There is little flexibility in this layout – people simply need to see out the front of their cars when driving. But once you remove this requirement, you can create a far more efficient layout. Seats can face inwards, allowing for easier interaction and collaboration with others, you can fit larger screens and equipment within, it would also not be out of the question to allow for sleeping room, or smaller overall cars without sacrificing interior space.

In the shared-vehicle system, space efficiencies will also extend to homes. Without having the need to have dedicated garages and carports, car maintenance tools, or as we will soon have – high-voltage electric car charging facilities, we’ll have more space within our homes itself – You know, to live in.

What has been done so far?

Some real-world applications then. The EUREKA Prometheus project (1987) was one of the first massively-funded R&D projects into the concept of an autonomous vehicle, achieving up to 158KM without human intervention. Since then, DARPA (the same one that invented the Internet and GPS, part of the US Department of Defence) has been one of the most prominent supporters of research into driverless cars, through events such as the DARPA Grand Challenge, a well-funded initiative to encourage innovations in autonomous vehicles.

One implementation of a self-driving car, is the BMW TrackTrainer – created in 2007. A pre-programmed road course (in this case either Laguna Seca, the Nordschleife or others) with the car driving around at it’s optimum, with the purpose to train the driver (or in the case, a passenger) on the best way to do a lap – using GPS data compared to an ‘ideal’ lap, assisted by a range of sensors around the car.

A more advanced implementation is Audi/VW‘s version of an automated TT-RS Pikes Peak car, designed to scale the famous mountain at a blistering pace (200km/h+) whilst being in a controlled slide during many of the corners. It has two main modules, one for safety algorithms and the other for vehicle dynamics – allowing the car to function at the absolute limit of traction. The GPS system also has an accuracy of two centimeters.

Testing an automated car in a close environment without other cars is a good start, no doubt this technology will filter down to street cars. But what about real traffic with real people? Arguably the most difficult challenge of all, Google has been at the forefront of self-driving automobile development on public roads, with it’s self-driving cars having covered more than 300,000 KM of real-world driving (with an average human intervention distance of 1,600KM) and having only one accident in total (it was human induced). This is no mean feat, as the infinite number of random variables on the roads can make automating a 1.5tonne vehicle effectively around difficult. Construction sites, other accidents, cyclists all pose major hazards to the vehicle while driving.

Each requires a huge array of sensors to effectively replace a human’s – In Google’s case, a Velodyne 64-beam laser, which scans and maps in a 360° area around the vehicle in full 3D detail (at no less than 1.3 million ‘points’ per second); 4 radars to detect objects in each direction at greater range; optical hazard recognition, GPS for locational data and a range of vehicle-movement sensors and gyroscopes. The idea is to feed all current environmental information (as well as pre-entered information) into the vehicle’s computer and allow it to travel along the road. Computers are also responsible for applying the required steering / throttle / brake inputs in order to safety traverse the environment, as well as handle any adverse situations – sudden icy roads, emergency braking/swerving manoeuvres and correcting any understeer/oversteer situations.

There are other technologies at play too that are already in use, such as widely-accepted Dedicated Short Range Communication (DSRC), which is a system of inter-vehicle communications, with the aim of reducing intersection crashes or enabling more cohesive cruise control / traffic management – it’s maximum range of 1KM doesn’t hurt either. More reading on supporting technologies. The most recent (and the most eye-catching) is BMW’s range of i3 and i8 electric / electric-hybrid cars, due in production by 2013. They utilise a series of cameras, radars and sensors to follow traffic up to 40km/h, providing steering, braking and throttle inputs, but they still do require one hand on the steering wheel. The Car-To-X system also provides advanced communication and warning with other vehicles, as pictured here.

What are the considerations?

There are of course, major hurdles to implementing these on a large scale. The technical obstacles, such as object detection, vehicle control and location finding, are rapidly being improved upon. However, there are other considerations before these cars become an integral part of our lives.

Legal implications – If an accident does occur, or somebody is injured in any automated car, who’s fault is it? Is it the manufacturer of the car or the owner of the car (whether it be the person, or a governmental / private body)? Is it the last mechanic that worked on the car? The bigger question is, just how much confidence will the population put in the failsafe’s of these vehicles? At least one US State has made it legal for self-driving cars to operate on their roads, but the general thoughts are that there is a lack of confidence in these systems.

What if an automated car collides with a human-controlled car, who’s fault is that? I can foresee insurance companies tackling these issues in the near future. Already, engineers have had to implement functions in these cars to work effectively with humans, such as creeping forward at intersections in order to signal intent.

Maintenance / Faults – Presumably, these cars with their multitude of systems will be next to impossible for an unqualified third-party to diagnose or repair. If a fault occurs with an indifferentiable model owned by a body, is it the body’s responsibility or the manufacturer’s responsibility?

Economics – Naturally, a huge number of industries will be affected (in both positive and negative ways) by the removal of the driver from the car. Mechanics, car accessory shops, aftermarket workshops, driver-assistance technologies, garage builders, to name a few. Government road authorities will need to be able to identify and modify current road networks as required, to optimise for these cars (such as dedicated autonomous vehicle lanes). Insurance companies will need to account for a vastly reduced number of accidents.

Non-mapped environments / emergencies – If you’re travelling off-road, unless that route has been pre-mapped, chances are the system will not work well. If the environment has recently changed or there has been an emergency (like heavy winds / flooding) which has changed the roads, the system will fail. The autonomous car, as we currently know it, only works in a pre-mapped environment, that has predictable obstacles and accessible roads. The question is then, what happens if you need to gain manual control over the vehicle? Will there be a provision to do so, and will that ability be abused by people who aim to gain some kind of advantage?

We’re still a solid 5-10 years away from commercially available completely self-driving cars, if not more. However, the benefits are clear and they are the future of our transportation (unless more accessible public transportation becomes the norm). Enthusiasts will change from lamenting the death of the manual transmission and internal combustion engines, to lamenting the death of driving as a daily part of our routine, but the writing is on the wall and we’ll all be better off for it.

Discussion on this post can be found here:

If you enjoyed this article, follow me on Google+ here


3 thoughts on “Self Driving Cars – Are They The Future?

  1. An excellent summary of the potential and huge advantages that this technology will eventually bring. One point perhaps not covered in any depth is the embedded energy in millions of cars that spend most of their life stationary and the poor use of assets and resources that this represents. Any road haulage or bus/coach business would be bankrupt in days if it made such poor use of its vehicles and other assets as we generally do of our cars. Also parked vehicles make unproductive use of countless kilomtres of existing road space as well as the aspect referred to of providing huge parking lots, individual parking spaces, garages and the rest of the existing ‘necessities’ that tie up valuable land area.
    The advantages of dispensing with individually owned and driven vehicles in favour of a public network of automated vehicles available on demand at short notice are obvious in urban situations but perhaps not so much so in more rural situations. However, given that the problems of congestion, pollution, space use and the general impact of transport on the environment are largely associated with conurbations anyway, perhaps this is not a critical factor?

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s