Lecture from the Rio-Campinas Connection partnership brings one of the most renowned researchers in the field of autonomous vehicles and transport
Cars are not computers yet, but they are getting closer. In the partnership's most recent live Rio-Campinas Connection this Wednesday (6), Canadian Brad Templeton gave a lecture on the future of this technology and what the next transport revolution should bring to life in cities. “Conexão”, a partnership initiative between the Department of Informatics (DI) and the Institute of Computing at Unicamp (IC), aims to disseminate the latest in the world of information technology through meetings with researchers in the area.
With a long trajectory in the development of autonomous vehicles, passing by companies like Google and Waymo, the researcher shared new discoveries, unprecedented projects and his extensive experience made throughout his career. Templeton introduced the talk by discussing a trend in computing: Moore's Law. According to this principle, the number of transistors on an integrated circuit (IC) doubles every two years, causing processors to increase their capacity exponentially, and the price even to drop. “This principle applies to almost all technologies, and with cars it's no different,” he said.
Putting the impact of the use of vehicles in cities in perspective, the researcher presented data that show the risks linked to driving by humans. Worldwide, 1,3 million people die in car accidents every year. This scenario implies a cost of 871 billion dollars in the United States alone. In addition to the human cost, Templeton also highlighted the huge environmental impact of these vehicles. According to him, there are 8 billion tons of carbon dioxide released into the atmosphere daily.
Today, although companies like Amazon and Baidu are at the forefront of autonomous vehicle development, the mass implementation of this technology is not really taking place yet. “Tesla, for example, does not have a fully autonomous car, but it does have autopilot options for semi-autonomy. This function still depends on the driver's attention. If you don't pay attention to the road, accidents can still happen”, added Templeton.
Seeking to change this scenario, Google has implemented a test model in a city in the suburb of Phoenix, capital of the US state of Arizona. There, fully autonomous cars, without the presence of control by a driver or member of the company, can be requested through an application. According to Templeton, this test never had a single failure or accident. “The future of mobility can guarantee a portable work environment. Without a driver, we can have seats vis-a-vis to other people and transform a fleet of cars into a kind of 'vehicle cloud', which are delivered on demand, wherever and whenever needed. These vehicles are still recharged and parked on their own”, he declared. And the passenger doesn't even have to worry about maintenance, cleaning or fuel, in addition to being able to occupy himself with other things while being transported”.
The researcher also explained that this autonomy does not only apply to cars, but also to planes and small aerial vehicles. Templeton highlighted that in São Paulo, one of the cities in the world with the greatest use of private helicopters to dodge traffic, autonomous air vehicles would be capable of transporting a considerable number of people and therefore would be a good solution for reducing traffic on land routes.
“The purpose of an entire city is transport. You want trips to be shorter to be able to meet and connect with other people, to get places quickly and comfortably. He also stressed that cities are always changing a lot with new types of transport. In the XNUMXth century, it was cars that allowed the creation of residential neighborhoods in the suburbs. So, with autonomous and “on demand” car fleets, other profound changes will certainly take place. The dynamics of the real estate market can also be changed. “In the United States, we say that the value of properties depends on three things: location, location and location. With the transformations in transport, this will also change.” I really hope that, in the future, parking lots will become parks,” he said.
At the end of his talk, Templeton opened the door for a question-and-answer chat with the attendees. Answering questions about the current failures of autonomous transport, the researcher did not mince any reservations when talking about the risks that still exist. “Computers are good at a lot of things, but humans are still better at predicting situations and acting as the scenario demands. If a person makes a mistake and gets a ticket, there is no guarantee that he will not do it again. But the robot, if he gets a single ticket, he'll never do it again. And better, all autonomous vehicles from that company will also no longer make this mistake Here is an advantage that a software update brings. And asked if autonomous cars use General Artificial Intelligence, he said no. They are able to learn, yes, but very specific things, for example recognizing pedestrians or making quick decisions to avoid accidents. “But humans still have an enormous capacity for abstraction that the machine does not have”, he concluded.
Missed this lecture? watch in our IC YouTube Channel.
Author: Corcovado Press Office