Autonomous cars have been subject of wishful thinking for ages; and these desires were even projected to television screen. Probably the biggest fun-favourite is the self-conscious four-wheel side-kick of David Hasselhoff from the 1980s, called Knight Industries Two Thousand, alias KITT. Among many other sci-fiesque functions, the wonder car was able to self-drive, give advice, make decisions, and even learn about human emotions and concepts.
No matter how high standers did KITT set for the future automobile-industry, present cars in some respect are way smarter than the technological star of the popular series. High-tech vehicles of today have at least one great advantage over the dream-car: they have internet, they are the inventions of IoT age. Even non-autonomous cars that are connected to internet via smartphones or any other way are able to indicate traffic jams, and construction works to warn us in time to change routs, which is by itself knowledge unimaginable for KITT.
Let’s see how far fiction is from reality. While in the imagination of the creators of KITT two front cameras, a front scanner, and a computer with 1000 megabytes of memory was sufficient for navigating on the roads of California without human interaction, there is way more data collection to it in reality. The latest Tesla model that has an autopilot function is equipped with a maximum 160-meter range radar, several side, front, and back cameras (the front camera having up to 250-meter range), and ultrasonics with maximum distance of 8 meters. The self-driving car project of Google called Waymo uses an at least 60-meter range LIDAR (a sensor rotating in 360° using laser to determine the position of surrounding objects), and internet as a basic source of information making use of IoT technology. The car is provided with an up-to-date map on which it positions all the relevant objects (cars, pedestrians, cyclists, policemen, road obstacles, and many more) in real time. This means that a smart car may “see” more than a human driver can on the road – as it also turns out from one of the examples presented by Chris Urmson, the former chief technology officer of Google’s autonomous car project during a 2015 TED Talk. He explained how the car spotted a cyclist from a safe distance who was just about to cross a red light, an observation humanly impossible but worth a life.
Notwithstanding, sheer recognition of the reckless cyclist, or as a matter of fact, any object participating in traffic is not enough for a fully functioning self-driving car: the vehicle needs to analyse the surrounding objects and predict their movement. The ability to predict was and still is a long learning process, and it is based on data accumulated during 2 million miles of test drives on real roads. Utilizing machine learning, the Google self-driving car has learnt what different types of vehicles, pedestrians, and cyclist look like and move, what possible and likely decisions they make, what signs and gestures mean on the road ranging from cyclist indicating a left turn to police officers guiding traffic; and also, that people are not always reliable. Chris Urmson also admitted in the TED Talk that there is always something new under the sun; even a program with 2 million miles of test run hasn’t seen everything. One of the unpredictable examples that Google car has encountered was a woman in an electric wheelchair chasing a duck in circles on the middle of the road. Urmson added that “it turns out, there is nowhere in the DMV [Department of Motor Vehicles] handbook that tells you how to deal with that.” This doesn’t mean that the car could not react properly in the situation: it has been taught to slow down and proceed safely even in such a peculiar situation.
This is where we can spot an important keyword, safety. As the fictional smart car, KITT was designed to protect human life, the (official) main objective of the development of smart cars is safety. The examples presented by Urmson indicate that
the least reliable part of the car” is the driver.
Statistics presented on Waymo website show an alarming number of 1.2 million annual deaths on the road, also adding that 94% of the accidents result from human errors.
Driving assistance (parking navigation, warning signals, etc.) is a big improvement in driver and passenger safety, but, according to the smart car guru, it cannot solve the problem by itself. While Tesla builds around the driver by assisting rather than replacing it, and granting the possibility to switch between automatic driving and driver control, former Google car, Waymo opts for complete autonomy of the car in the long run. Urmson argues that test driving experiences have shown that assisted drivers are less cautious, which is a major risk factor on the road. Guardian came to the same conclusion using the analogy of the autopilots of the flying industry, where combining auto- and real pilot does not necessarily give the best results. Pilots flying with autopilot mode have less experience in actual control of the plane (as they primarily supervise the autopilot) which comes as a significant disadvantage in a crisis. Besides, reliance on automatization often decreases alertness, while often there is confusion about when it is the right time to take over from the automatic control.
The main question is what the numbers say about the safety of autonomous cars, especially after the first fatal accident happened in 2016 involving a Tesla in autonomous mode, where, at the end, the company was not found responsible. According to the other smart car guru’s own reports and statistics, during the 2 million miles spent on real traffic Waymo or Google self-driving car was involved in only couple of minor accidents; and Google took partial responsibility in only one case (where the car made the same decision as the driver would do). Unfortunately, the Google project rebranded as Waymo no longer publishes its monthly accident reports, but as far as US cases are concerned self-driving car accidents can still be followed on the Department of Motor Vehicles website. Moreover, a larger scale comparison between regular and self-driving car safety performance also came to light. Safer America compared the available statistics on the accidents including human drivers and self-driving cars concluding that while one fatality occurs with an individual behind the wheel for every 90 million miles driven, this number is 130 million miles in cars with autocontrol. However, it is important to note that the data are not representative, as there is substantially less information available about self-driving cars, which are still only in test mode. This means that future is yet to determine the winner of the safety race.
In addition to the protection of human life, an argument supporting autonomous cars is energy saving, as the car can adjust to use less fuel or electricity in traffic. This means improvement regarding cost efficiency and environment protection as well. If regulations allow fully self-driving cars to hit the roads, it can grant more flexibility to people with disabilities who are otherwise not able to drive. Reducing time spent with driving can also free up extra productive hours for daily commuters.
Nonetheless, it is not only individuals who can see the daily advantages of the cross section of IoT and automotive industry. Some cities, such as Lausanne, Wageningen and Trikala have already welcomed driverless vehicles. Environmental friendliness, cost efficiency due to travel optimization and reduction of labor costs, and the often-mentioned safety are the main reasons why municipalities opt for adopting the new technology.
However, saying goodbye to the driver is not the only way to modernize public transport via IoT. An approach called fleet telematics can ensure the best use of the available fleet. Cloud connected passenger counters can monitor the traffic and determine vehicle needs in certain routes during peak and off-peak hours to optimize schedule. Information combined with tracking the position of the available vehicles makes flexible, demand adjusted, real time scheduling possible. In addition to passenger counters, vehicle and driver performance can be easily followed with tracking. Sensors can also send up-to date data about maintenance needs and fuel or electricity consumption. Consequently, IoT can reduce expenses by cutting maintenance costs and maximizing the utilization of the available fleet by avoiding half-empty journeys. Moreover, it improves customer satisfaction by avoiding overcrowded vehicles and providing feedback about drivers.
As Red Hat explains in one of its case studies, IoT has its fair share in making train journeys safer. During the North American Positive Train Control (PTC) program, the company plans to install 60,000 devices to 70,000 miles of track and 25,000 locomotives. The rail-side sensors were designed to follow train speed and load. On-board control and monitoring systems provide maintenance information and include automatic breaking system to avoid any accidents. To avert collisions, derailments, and any interference, the IoT system follows the position of the trains in real time and guides them while dynamically re-routes train traffic if necessary, and takes action with automatic breaks if accidents are about to happen. The applied technology aids to keep the schedule, which is necessary for customer satisfaction. It also helps to meet the valid governmental and industry regulations as train speed, condition and load are constantly monitored and controlled. Needless to say, it also reduces maintenance and fuel consumption costs while keeping safety as a top priority.
Such kinds of solutions are not exclusive to public transport. Any company which needs to ship, and deliver products can make use of connected vehicles. Scheduling and planning loading, placement, routs, and delivery can be optimized and automatized by IoT based logistics. Products can reach their destination on the most effective and safest routs without wasting the capacity of the carefully chosen vehicle saving time and money. Smart monitoring systems are especially useful when sensitive cargo is about to be shipped. Such factors as temperature, light, humidity and vibration can be measured in real-time and alarms can be created to warn the driver if actions need to be taken for human or product safety when transporting fragile, sensitive goods, and hazardous materials. Live observation of the cargo also facilitates theft prevention. For industrial fleets, there is also lot to gain including reduction of maintenance costs, improving safety and customer satisfaction, providing faster services.
In addition to individuals and industry, innovative governments also make use of traffic related IoT technology for the benefit of the travelers, environment, and economy. As it turns out from another case study, European transportation ministry in collaboration with Red Hat put smart toll collection in action in 14,000 miles of multinational highway network. The project aims to implement a stoppage and delay free toll collection system for commercial vehicles, where fees are calculated based on weight and vehicle class. The system is supplemented with a customer-facing user interface that enables flawless payment. Red Hat explained in the case study that the system required the implementation of truck transporters, GPS sensors, roadside beacons and signals, and a carefully connected chain of sub-systems. After precise and reliable data are gathered by the sensors and the combined data are transmitted right to financial institutions. The project faced and tackled the challenge of providing a multinational service integrating several systems while maintaining data precision and privacy safety. As a reward the working system ensures flawless traffic that is advantageous not only for consumer convenience but environmental friendly as well. It eliminated the risks of cash handling, and reduced labor and operation expenses.
After all the diverse application possibilities, it comes as no surprise that transportation has one of the biggest shares in IoT market. According to statistics published by Forbes, global IoT spending is expected to reach 250 billion euros in 2020 from which transportation alone will cover 40 billion euros similarly to discrete manufacturing. These two fields came first in the 2015 buyer survey as well with 10 billion euros each. No wonder why the Forrester’s The Internet of Things Heat Map, 2016 marked several aspects of transportation IoT as the “hottest” on the rapidly increasing field including opportunities in supply chain and fleet management. GSM Association goes as far as suggesting that every car will be linked to the internet by 2025.
All signs indicate that it is just the right time to find a customized, smart way to get connected.