The Internet of Things in Transportation: Not So Fast

Is the Internet of Things (IoT) the Next Big Thing for transportation, or does it mean opening up your car or bus or train to hackers? Can a hacker detune your engine, suddenly reverse the action of your steering wheel, make the airbags deploy when you hit 110 kilometer per hour, reprogram your anti-lock brake system, or just disable the whole vehicle?

Without well thought-out security, IoT is worthless. At present, there are several different “standards” for IoT, so the concept does not yet really exist.

The problem with the Internet of Things: Complexity

The core problem is complexity. Any system as complex as the Internet is going to have unanticipated exploits, which is why desk-top systems need to get constantly updated. On a vehicle, malicious software can literally kill you.

Vehicle components (for example, engine control unit, door locks, air bags) are simple and contain relatively few lines of code. At this size it is feasible to use formal methods to mathematically prove that software correctly implements its specifications. With larger systems, this becomes impossible.

Some argue that we need IoT so that we can instantly upgrade software. A better approach may be to not let IoT touch the system in the first place; take the vehicle to the dealer if the upgrade is important. Even a benign upgrade can cause problems if the driver is unaware that her controls now behave differently.

The world’s militaries are actively working on cyber-warfare. The IoT may enable them to shut down an entire country’s transportation system.

The technical side of the Internet of Things

A modern automobile may have 50 or more microprocessors, with nobody in charge. Safety critical functions must be kept separate from infotainment items; ideally by keeping them on two physically separated wiring systems.

A commonly used system is the CAN bus, which replaces bulky wiring harnesses with four wires to all components: two signal lines, power and ground. Any component can put a message on the bus, perhaps at the same time, and a clever scheme arbitrates them according to priority.

The Internet has a similar architecture of nobody in charge, but it commonly has wireless links to remote sites. In either CAN bus or IoT, a component has the responsibility to read a message, authenticate it, and decide whether it is relevant.

The Internet of Things: Connectivity challenges

A vehicle needs to connect to the outside world for infotainment, navigation assistance, and communication with other vehicles and the infrastructure. This requires a strong firewall between outside connections and the vehicle control systems.

With automated driving, the current firewall between vehicle control and infotainment might be selectively reduced. If the CAN bus is broadcast, a following vehicle can detect braking before the brake lights go on. An external computer might control a platoon of vehicles at the level of taking over their accelerators, brakes and steering.

Access to vehicle control needs to be secure. Some vehicle designers advocate relying on a central computer that operates everything else.

Would security on a centralized system be better or worse? How can we have confidence that our wireless links do not expose vehicle control to hackers? Share your thoughts in the comment section.

Please note that this article expresses the opinions of the author and does not reflect the views of Move Forward.

How Automated Vehicles Reduce Congestion

Autonomous vehicles promise reduced road accidents and an improved rider experience. Autonomy does nothing for congestion, but electronically connecting the vehicles does.

An autonomous vehicle need not be connected; fancy instrumentation and software may replace the human driver. Communicating with others reduces the technical problems. Costs tumble because vehicles do not need the most expensive equipment; sharing data improves each vehicle’s estimated relative position.

Many new cars offer Adaptive Cruise Control (ACC), a feature that tracks the position of the car ahead, and varies vehicle speed as appropriate. ACC without communication produces the same stop and go as manual drive. Supplementing ACC with information about several leading vehicles yields smooth platoon acceleration or deceleration; this technology by itself can double freeway lane capacity.

Data to share include each vehicle’s best estimate of its position, heading and speed, as well as its estimated distances to adjacent vehicles. These data can be uploaded to the cloud, which can look at the overall situation, improve and download the individual estimates.

Better road utilization

All vehicles sold in Europe and North America are required to have on-board diagnostics. The diagnostic system knows everything happening in the vehicle – including state of the accelerator, brakes, and steering wheel. These data can be shared with nearby vehicles.

Manually driven vehicles need to maintain minimum spacing so that the driver can observe and react to unexpected behaviors. Automated vehicles can operate with a much smaller gap, especially if they foresee the intended actions of the car ahead.

Automated vehicles travel in platoons, with distances between vehicles as small as three meters at freeway speed. Highway capacity increases by a factor of three. An external traffic computer can control vehicle behavior.

Road traffic: The end of Stop and Go

When manually driven vehicles are stopped at a red light, starting again is a chain reaction. Each driver needs to verify that the car ahead has started moving, and that there is a reasonable gap. Starting a line of stopped vehicles is a slow process.

By contrast, a stopped automated platoon has no such restriction. Every vehicle can start moving at the same time.

Driver slowing reaction to a hill or a roadside distraction produces a shock wave that slows following vehicles. Automated control smoothens or eliminates such slow-downs. Vehicles within a platoon are tightly spaced, but a large gap to the next platoon defeats shock waves.

Some people expect that we will perfect autonomy, then at a later stage, get the vehicles to cooperate. A better course is to set up cooperation first on dedicated lanes, bypassing the technically difficult problem of designing a vehicle that can deal with any situation.

Security: A top priority for autonomous vehicle technology

Sharing data between vehicles is advantageous, but carries the danger of exposure to hackers. Someone could mess with your Engine Control Unit or change the driving instructions that were meant to keep you in the platoon. If the industry fails to adequately address security, automated vehicle technology could die.

If done right, we can have more road capacity without pouring any more concrete. However, strong security is essential.

What would convince you to trust your car to an external road traffic control computer? Do the promises of reduced congestion merit increased government spending to implement automated transportation? Share your thoughts in the comment section.

Please note that this article expresses the opinions of the author and does not reflect the views of Move Forward.

How Does the Fully Automated Vehicle Get on the Road?

When will it be safe to turn loose the first fully automatic car? Is there any amount of testing that can prove that the technology is ready to be deployed?

Automobile safety is a paradox: Globally, traffic accidents kill 1.2 million people annually, with 90 percent of fatalities in the developing world, which has less than half of the world’s vehicles. Yet in the U.S. fatal crashes only happen once in 100 Million miles (160 M km) of travel. Can self-driving cars do just as well?

There are several scenarios that could put full automation on the road.

• Test the system with a standby driver for a million or more kilometers and then release the cars to the highway. But in a million km of driving, there are some truly baffling situations that happen rarely. How safe does an automated system need to be?

• Focus on a low speed system that can deal with pedestrians and bicyclists. The European City Project demonstrating automated road passenger transport follows this approach.

• Simplify the problem by separating automated vehicles from pedestrians and manually driven vehicles. This is the method for Personal Rapid Transit (PRT) and automated passenger trains. Such systems have been operational for 40 years.

What is the purpose of self-driving vehicles?

Is it only to provide convenience and safety? If we also want to cut traffic congestion, we need to examine a wider context than the individual vehicle.

Automated cars can be choreographed to smoothly flow in platoons. If the automated lane is moving nicely, but the manual lanes are only creeping, manual drivers would move into the automated lane and snarl it. Thus reducing congestion requires separate lanes, which puts self-driving vehicles into the PRT model.

When a vehicle only has to deal with other automated vehicles, it requires less intelligence and fewer sensors, bringing the cost down.

Putting automated vehicles in a reserved lane presents a chicken and egg problem: there are not enough automated vehicles to justify a separate lane and there are few automated vehicles because there is no place to drive them. The solution is the rail transit approach: acquire the vehicles and their guide ways at the same time.

The public transit model may be more appropriate than the private car model for introducing automation. Once a public system is in place, it could expand to accept properly qualified private vehicles.

It is not a question of waiting for the technology to mature. The mobility problem can be constrained to match today’s technology. We need forward thinking individuals and governments willing to go beyond transit as it has always been done.

Share your opinion in our comment section: Which model is best suited to introduce automation – the public transit or the private vehicle and why?

Please note that this article expresses the opinions of the author and does not reflect the views of Move Forward.

When Will Automated Vehicles Become a Reality?

Driverless cars are in the active experimental phase, with a single company amassing more than two million autonomous kilometers. A common question is: “When will the first autonomous vehicles be deployed in the real world?”

The answer is 1924, which saw the first elevator that did not require an operator. Elevators are not inherently autonomous. Our thinking evolves as the technology becomes commonplace.

Maybe the question is: “When will the first driverless automated horizontal land vehicles start public operation?”

The answer is 1975. The Morgantown, WV USA Group Rapid Transit system has been operating for 40 years, carrying 16,000 riders a day. There are 130 other driverless rail systems operating today. These include the metro system in Lille, France, driverless since 1983, Vancouver’s Skytrain system since 1985 and shuttle systems at numerous airports.

Rephrasing the question again: “Will the first driverless road vehicle using rubber wheels on a paved roadway at grade level become operational this century?”

The answer is “No”. Since 1999, the city of Rivium in the Netherlands has operated a driverless mini-bus on a separated busway with top speed of 32 km/h. Smaller car-sized vehicles are operating on dedicated paved paths at London’s Heathrow airport and in the eco-city of Masdar, UAE.

Practical land vehicle automation today

The secret to making land vehicle automation practical today is to not mix automated vehicles with manually driven vehicles. Doing so makes the technology much more complicated. If the objective is to cut congestion, you don’t want to mix them anyway.

Final question: When will ordinary people be able to operate automobiles on public highways with no driver intervention?

The answer is 2017. The Swedish government has partnered with a car company to turn 100 cars loose on the freeways of Gottenberg.

At least two manufacturers are promising self-driving cars by 2020. However, there are already cars operating on the roads with all the driver assistance features needed for autonomous operation.

Automation is inevitable. The real question is not if or when, but what form will automation take?

Have you had experiences with any of the systems mentioned? How was it?

Please note that this article expresses the opinions of the author and does not reflect the views of Move Forward.

Automated Vehicle Does Not Mean Car

Automated driving can change the very notion of what car means, especially in the city. The pod of the future may have little resemblance to an automobile.

Globally, traffic accidents kill 1,240,000 people annually. Driver error is the factor in over 90 percent of accidents. In the USA, alcohol is involved in 31 percent of fatal accidents.

Automated road transportation offers hope of reducing the carnage. The technology will never make it to the road until it has matured to be safer than human drivers.

The victim is often not the driver. Pedestrians, bicyclists and motorcyclists die in disproportionally high numbers. People in an SUV are three times more likely to survive a head-on collision than those in a light car.

If accidents do become rare, it will be almost as safe to ride a motorcycle as an SUV. The driver may not see the cyclist, but the sensors will, and will not allow a collision.

The Economics of Automation

Current prices of advanced driver assistance systems (ADAS) limit them to luxury cars. However, pricing for electronics follows an unstoppable downward path. A computer comparable to one that you can buy today for $30 would have cost $300,000 forty years ago, without adjusting for inflation.

Automotive RADAR or LIDAR units currently cost thousands of Euros, but in a few years we may see units costing under 100 Euros. Stereo cameras in this low price range can do the same job. It should be possible to build complete electronics for a vehicle (computer, sensors and actuators) for 1000 Euros.

Solution for reducing traffic congestion

An autonomous car may be safe and convenient, but individual automation has no effect on congestion. Congestion can be reduced by externally choreographing all road users and operating them so that they do not bunch up.

With inexpensive equipment, cyclists can take their rightful place in traffic. A cyclist could activate a smartphone app that broadcasts her position to other vehicles and makes her part of the traffic stream. There may be dynamic bicycle lanes.

Another option is to modify the bicycle so that it can be externally controlled. By adding a third wheel for balance, an electric motor, and actuators for steering and brakes, bicycles can move as fast as cars in the city.

With a shell for weather protection, the new urban vehicle of choice may be an overgrown bicycle that can run from pedal power, where the pedals are used to charge a battery. The drive itself would be electric, so that the bike can be externally controlled and flow with the rest of traffic.

This is not the bike to take on a recreational ride, but it is a great way to get to work. Similarly, automobile aficionados can take a Sunday drive in the country. Does anyone really enjoy creeping through city traffic for an hour?

Self-drive can take us to a post-automotive age. Don’t assume business as usual.

Share your opinion about the concept of automated bicycles

Please note that this article expresses the opinions of the author and does not reflect the views of Move Forward.