Just like teenagers, self-driving cars need practice to really learn to drive

A self-driving car, of unknown level of education. Image: Grendelkhan/Flickr/creative commons.

What do self-driving cars and teenage drivers have in common?

Experience. Or, more accurately, a lack of experience.

Teenage drivers – novice drivers of any age, actually – begin with little knowledge of how to actually operate a car’s controls, and how to handle various quirks of the rules of the road. In North America, their first step in learning typically consists of fundamental instruction conveyed by a teacher. With classroom education, novice drivers are, in effect, programmed with knowledge of traffic laws and other basics. They then learn to operate a motor vehicle by applying that programming and progressively encountering a vast range of possibilities on actual roadways. Along the way, feedback they receive – from others in the vehicle as well as the actual experience of driving – helps them determine how best to react and function safely.

The same is true for autonomous vehicles. They are first programmed with basic knowledge. Red means stop; green means go, and so on. Then, through a form of artificial intelligence known as machine learning, self-driving autos draw from both accumulated experiences and continual feedback to detect patterns, adapt to circumstances, make decisions and improve performance.

For both humans and machines, more driving will ideally lead to better driving. And in each case, establishing mastery takes a long time. Especially as each learns to address the unique situations that are hard to anticipate without experience – a falling tree, a flash flood, a ball bouncing into the street, or some other sudden event. Testing, in both controlled and actual environments, is critical to building know-how. The more miles that driverless cars travel, the more quickly their safety improves. And improved safety performance will influence public acceptance of self-driving car deployment – an area in which I specialise.

Starting with basic skills

Experience, of course, must be built upon a foundation of rudimentary abilities – starting with vision. Meeting that essential requirement is straightforward for most humans, even those who may require the aid of glasses or contact lenses. For driverless cars, however, the ability to see is an immensely complex process involving multiple sensors and other technological elements:

  • radar, which uses radio waves to measure distances between the car and obstacles around it;
  • LIDAR, which uses laser sensors to build a 360-degree image of the car’s surroundings;
  • cameras, to detect people, lights, signs and other objects;
  • satellites, to enable GPS, global positioning systems that can pinpoint locations;
  • digital maps, which help to determine and modify routes the car will take;
  • a computer, which processes all the information, recognising objects, analysing the driving situation and determining actions based on what the car sees.

How a driverless car ‘sees’ the road.

All of these elements work together to help the car know where it is at all times, and where everything else is in relation to it. Despite the precision of these systems, however, they’re not perfect. The computer can know which pictures and sensory inputs deserve its attention, and how to correctly respond, but experience only comes from traveling a lot of miles.

The learning that is occurring by autonomous cars currently being tested on public roads feeds back into central systems that make all of a company’s cars better drivers. But even adding up all the on-road miles currently being driven by all autonomous vehicles in the U.S. doesn’t get close to the number of miles driven by humans every single day.

Dangerous after dark

Seeing at night is more challenging than during the daytime – for self-driving cars as well as for human drivers. Contrast is reduced in dark conditions, and objects – whether animate or inanimate – are more difficult to distinguish from their surroundings. In that regard, a human’s eyes and a driverless car’s cameras suffer the same impairment – unlike radar and LIDAR, which don’t need sunlight, streetlights or other lighting.

This was a factor in March in Arizona, when a pedestrian pushing her bicycle across the street at night was struck and killed by a self-driving Uber vehicle. Emergency braking, disabled at the time of the crash, was one issue. The car’s sensors were another issue, having identified the pedestrian as a vehicle first, and then as a bicycle. That’s an important distinction, because a self-driving car’s judgments and actions rely upon accurate identifications. For instance, it would expect another vehicle to move more quickly out of its path than a person walking.


Try and try again

To become better drivers, self-driving cars need not only more and better technological tools, but also something far more fundamental: practice. Just like human drivers, robot drivers won’t get better at dealing with darkness, fog and slippery road conditions without experience.

Testing on controlled roads is a first step to broad deployment of driverless vehicles on public streets. The Texas Automated Vehicle Proving Grounds Partnership, involving the Texas A&M Transportation Institute, University of Texas at Austin, and Southwest Research Institute in San Antonio, Texas, operates a group of closed-course test sites.

Self-driving cars also need to experience real-world conditions, so the Partnership includes seven urban regions in Texas where equipment can be tested on public roads. And, in a separate venture in July, self-driving startup Drive.ai began testing its own vehicles on limited routes in Frisco, north of Dallas.

These testing efforts are essential to ensuring that self-driving technologies are as foolproof as possible before their widespread introduction on public roadways. In other words, the technology needs time to learn. Think of it as driver education for driverless cars.

People learn by doing, and they learn best by doing repeatedly. Whether the pursuit involves a musical instrument, an athletic activity or operating a motor vehicle, individuals build proficiency through practice.

The ConversationSelf-driving cars, as researchers are finding, are no different from teens who need to build up experience before becoming reliably safe drivers. But at least the cars won’t have to learn every single thing for themselves – instead, they’ll talk to each other and share a pool of experience.

Johanna Zmud, Senior Research Scientist, Texas A&M Transportation Institute, Texas A&M University .

This article was originally published on The Conversation. Read the original article.

 
 
 
 

The risk of ‘cascading’ natural disasters is rising

A man watches wildfires in California, 2013. Image: Getty.

In a warming world, the dangers from natural disasters are changing. In a recent commentary, we identified a number of costly and deadly catastrophes that point to an increase in the risk of “cascading” events – ones that intensify the impacts of natural hazards and turn them into disasters.

Multiple hazardous events are considered cascading when they act as a series of toppling dominoes, such as flooding and landslides that occur after rain over wildfires. Cascading events may begin in small areas but can intensify and spread to influence larger areas.

This rising risk means decision-makers, urban planners and risk analysts, civil engineers like us and other stakeholders need to invest more time and effort in tracking connections between natural hazards, including hurricanes, wildfires, extreme rainfall, snowmelt, debris flow, and drought, under a changing climate.

Cascading disasters

Since 1980 to January 2018, natural disasters caused an inflation-adjusted $1,537.4bn in damages in the United States.

The loss of life in that period – nearly 10,000 deaths – has been mounting as well. The United States has seen more billion-dollar natural disaster events recently than ever before, with climate models projecting an increase in intensity and frequency of these events in the future. In 2017 alone, natural disasters resulted in $306bn losses, setting the costliest disaster year on record.

We decided it was important to better understand cascading and compound disasters because the impacts of climate change can often lead to coupled events instead of isolated ones. The United Nations Office for Disaster Risk Reduction, or UNISDR, claims: “Any disaster entails a potentially compounding process, whereby one event precipitates another.”

For example, deforestation and flooding often occur together. When vegetation is removed, top soil washes away and the earth is incapable of absorbing rainfall. The 2004 Haiti flood that killed more than 800 people and left many missing is an example of this type of cascading event. The citizens of the poverty-stricken country destroyed more than 98 per cent of its forests to provide charcoal for cooking. When Tropical Storm Jeanne hit, there was no way for the soil to absorb the rainfall. To further complicate existing issues, trees excrete water vapor into the air, and so a sparser tree cover often yields less rain. As a result, the water table may drop, making farming, which is the backbone of Haiti’s economy, more challenging.


Rising risk from climate change

Coupled weather events are becoming more common and severe as the earth warms. Droughts and heatwaves are a coupled result of global warming. As droughts lead to dry soils, the surface warms since the sun’s heat cannot be released as evaporation. In the United States, week-long heatwaves that occur simultaneously with periods of drought are twice as likely to happen now as in the 1970s.

Also, the severity of these cascading weather events worsens in a warming world. Drought-stricken areas become more vulnerable to wildfires. And snow and ice are melting earlier, which is altering the timing of runoff. This has a direct relationship with the fact that the fire season across the globe has extended by 20 per cent since the 1980s. Earlier snowmelt increases the chance of low flows in the dry season and can make forests and vegetation more vulnerable to fires.

These links spread further as wildfires occur at elevations never imagined before. As fires destroy the forest canopy on high mountain ranges, the way snow accumulates is altered. Snow melts faster since soot deposited on the snow absorbs heat. Similarly, as drought dust is released, snow melts at a higher rate as has been seen in the Upper Colorado River Basin.

Fluctuations in temperature and other climatic patterns can harm or challenge the already crumbling infrastructure in the United States: the average age of the nation’s dams and levees is over 50 years. The deisgn of these aging systems did not account for the effects of cascading events and changes in the patterns of extreme events due to climate change. What might normally be a minor event can become a major cause for concern such as when an unexpected amount of melt water triggers debris flows over burned land.

There are several other examples of cascading disasters. In July, a deadly wildfire raged through Athens killing 99 people. During the same month on the other side of the world in Mendocino, California, more than 1,800 square kilometers were scorched. For scale, this area is larger than the entire city of Los Angeles.

When landscapes are charred during wildfires, they become more vulnerable to landslides and flooding. In January of this year, a debris flow event in Montecito, California killed 21 people and injured more than 160. Just one month before the landslide, the soil on the town’s steep slopes were destabilised in a wildfire. After a storm brought torrential downpours, a 5-meter high wave of mud, tree branches and boulders swept down the slopes and into people’s homes.

Hurricanes also can trigger cascading hazards over large areas. For example, significant damages to trees and loss of vegetation due to a hurricane increase the chance of landslides and flooding, as reported in Japan in 2004.

Future steps

Most research and practical risk studies focus on estimating the likelihood of different individual extreme events such as hurricanes, floods and droughts. It is often difficult to describe the risk of interconnected events especially when the events are not physically dependent. For example, two physically independent events, such as wildfire and next season’s rainfall, are related only by how fire later raises the chances of landslide and flooding.

As civil engineers, we see a need to be able to better understand the overall severity of these cascading disasters and their impacts on communities and the built environment. The need is more pronounced considering the fact that much of the nation’s critical infrastructure is aged and currently operate under rather marginal conditions.

A first step in solving the problem is gaining a better understanding of how severe these cascading events can be and the relationship each occurrence has with one another. We also need reliable methods for risk assessment. And a universal framework for addressing cascading disasters still needs to be developed.

A global system that can predict the interactions between natural and built environments could save millions of lives and billions of dollars. Most importantly, community outreach and public education must be prioritised, to raise awareness of the potential risks cascading hazards can cause.

The Conversation

Farshid Vahedifard, CEE Advisory Board Endowed Professor and Associate Professor of Civil and Environmental Engineering, Mississippi State University and Amir AghaKouchak, Associate Professor of Civil & Environmental Engineering, University of California, Irvine.

This article is republished from The Conversation under a Creative Commons license. Read the original article.