Situational awareness key to safe, self-driving cars

This is another in a series entitled The Future of Mobility, a joint project of CommonWealth and Meeting of the Minds, a San Francisco-based organization that seeks to build alliances around urban sustainability.

THREE TECHNOLOGIES INTRODUCED in the first half of the 20th Century have completely transformed the way all Americans live today relative to our predecessors 100 years ago: the automobile, the transistor, and wireless communication.

The mass production of the automobile has made us a highly mobile society and workforce, ushered in the age of the suburb (and the slow decay of many urban centers), and changed national commerce by allowing for the rapid transportation of goods across the country.

The transistor, which signaled the start of the Information Age when introduced in 1947, is the fundamental ingredient for all digital microelectronics and microprocessor systems, which are at the core of the electronic devices that are so ubiquitous today, from laptops, computers, and smart phones to refrigerators, cameras, and TVs.  Furthermore, since the implementation of the ARPANet and its evolution into the Internet, more and more of these electronic appliances have become connected, giving rise to the so-called “Internet of Things.”

Since 1901, when Guglielmo Marconi demonstrated the feasibility of long-range radio transmission between Poldhu, in Cornwall, England, and Signal Hill in St. John’s, Newfoundland, wireless connectivity has become another enabler and vital component of the Information Age, significantly impacting commercial, educational, national defense, entertainment, and political activities.

As these technologies have continued to evolve into applications that enhance our daily lives, they have also cross-pollenated with each other in some ways.  Focusing on automotive technology, all auto manufacturers started incorporating specialized microprocessing systems, called electronic control units, throughout vehicles to enable functions ranging from fuel-injection, power steering, and automatic transmission to features such as powered windows, powered door locks, and integrated audio systems.  At present, vehicles have  approximately 100 electronic control units, which are all connected to each other via an onboard internal communication network referred to as a controller area network bus. This connectivity allows for optimization of fuel efficiency, performance, and/or driver safety.

One application receiving a substantial amount of attention over the past few years is autonomous vehicles, or self-driving cars.  These automotive systems promise to enhance safety relative to human-operated vehicles.  By enabling these vehicles to make driving decisions in real-time and across a range of real-world scenarios with the primary goal of maximizing safety, this application has become quite appealing to a society where approximately 90 driving-related fatalities occur per day in the US alone (based on 2014 statistics).  For self-driving automobiles to succeed at maximizing safety, they need extensive digitization of their onboard operations; reliable and real-time decision-making algorithms, many of which have been developed by the artificial intelligence research community; and extensive situational awareness of the vehicular operational environment.

Situational awareness, or the full understanding of the surrounding environment, is critical for safety.  Although camera/image-based systems (LIDAR, RADAR, and other sensors) currently provide some information about the operating environment, they don’t provide the “whole picture” about what is happening around the vehicle, and decisions based on this incomplete information can lead to fatal consequences. Gathering situation awareness data remains a significant technical challenge for the automotive sector and is, so far, a barrier to developing vehicles that are reliable, safe, and ready for widespread adoption by the consumer automobile market: How can autonomous vehicles read their full environment, as humans do?

One answer may lie in wireless communications. Since the mid-1990s, wireless connectivity technology has been used in vehicles for services such as OnStar and GPS navigation systems, and in e-tolling through RFID technology. In the early 2000s, Bluetooth technology became a standard feature in most vehicle audio systems, while wireless tire-pressure monitoring system devices have been mandated in all US vehicles since 2008.

Through wireless communications, vehicles can connect and communicate with each other, sharing their situational awareness information. Called “connected vehicles” or “vehicle-to-vehicle,” this type of communication is a game-changer, especially for self-driving cars.  Instead of each vehicle receiving information only from its own sensors, vehicles using connected vehicle technology can gather a network of information regarding events or actions that might occur beyond any one particular vehicle. This comprehensive situational awareness, referred to as a “vehicular cloud,” even exceeds the capabilities of all human operators, thus meeting society’s expectations of reliability and safety.

The technology is promising but relies on dedicated wireless channel bandwidth. In 1999 the Federal Communications Commission allocated 75 MHz of wireless channels around 5.9 GHz; this is the only wireless channel bandwidth currently dedicated to vehicular communications anywhere in the US.

Bandwidth is a finite resource and therefore subject to fierce competition among industries. With the exponential growth in smart phones and WiFi-enabled laptops/devices, including household appliances, industries must compete for bandwidth. Because the automotive sector has thus far underutilized its 75MHz—due in part to the slow development of the needed technologies— other sectors are pushing to gain access to it. Government proceedings are currently underway with the FCC, and connected vehicle applications may be required to share their channels with cellular phone, WiFi, and other non-automotive wireless applications.

This dedicated channel is critical to the safety of self-driving vehicles: Without it, we lose all ability to support vehicular connectivity, as the formation of vehicular clouds across our nation’s roads and highways would be jeopardized.  The loss of dedicated wireless channel bandwidth for connected vehicle communications would negatively impact the automotive sector, and the revolution in self-driving car technology would be dealt a significant blow.  It would impact the reliability and safety of self-driving cars, especially during life-threatening vehicle situations.  Alternative forms of wireless communications suggested for connected vehicles, such as WiFi and advanced cellular telephony, are insufficient –and dangerous—because of delays associated with their transmitter and receiver/decoder technologies. Technologies that support sharing the bandwidth—referred to as Dynamic Spectrum Access—have not evolved sufficiently yet to avoid communication conflicts with each other. (In 2009, for instance, my research laboratory studied the performance of vehicle-to-vehicle across unused digital television channels.  Although promising, rules for the sharing of these wireless channels with other applications such as IEEE 802.22 Wireless Regional Area Networks and 4G LTE are relatively unclear at best.)

The FCC and others will soon announce a decision regarding the 75MHz in question. It remains to be seen whether greater cell phone access and convenience will be obtained at the expense of automobile safety.

Alexander M. Wyglinski is associate professor of electrical and computer engineering at Worcester Polytechnic Institute. He is also the president of the Institute of Electrical and Electronic Engineers Vehicular Technology Society.