The adoption of autonomous vehicles on a global scale is picking up speed. The United Kingdom has recently passed the Automated Vehicles Act in order to establish the safe integration of fully and partially autonomous vehicles into society over the next few years. More autonomous vehicles are being tested right now in China than anywhere else in the world. And in the United States, major metropolitan areas have enlisted the use of “robotaxis” in their public transportation capabilities. Companies like Cruise, Waymo and, of course, Tesla all have billions of dollars invested in their grand ambitions of hosting self-driving cars and services from coast-to-coast and all over the world. At this point, the development and implementation of autonomous vehicle technology is no longer a matter of ‘if’ or really even ‘when’, but simply a matter of ‘to what extent’? What can the adoption of AI-powered autonomous vehicles on a massive scale do to improve not just our roadways but our society?
Automating Road Safety
The push for autonomous vehicles and the mountains of capital invested in these technologies is indicative of the widely acknowledged public good that the deployment of self-driving cars can have. For starters, there are the safety standards of self-driving vehicles over that of human drivers. A recent study published in Nature Communications and insights from Tesla’s 2022 Impact Report underscore the transformative potential of autonomous vehicle implementation on the enhancement of road safety. AVs have been found to reduce rear-end, head-on and lateral collisions, as well as incidents of running off the road, by 20% to 50%. Given that the World Health Organization estimates that road traffic injuries are responsible for the death of 1.35 million people each year across the world, this dramatic improvement in automobile safety would have a seismic impact. While more technological refinement is still required before autonomous vehicles outperform human drivers in all circumstances (human-driven cars still remain safer in low-visibility conditions and during turns), the further advancement of sensor technologies, predictive algorithms and V2X communications will continue to improve responses in these complex driving scenarios and enable them to create safer roadways.
AV-oiding Traffic
In addition, the deployment of autonomous vehicles will also have a significant impact on the issue of traffic congestion. In a study conducted by the Association for Commuter Transportation (ACT) and the United States Department of Transportation (USDOT), “rush hour” commutes—once an appropriate title—now make up six hours a day and make travel during these heightened times take 40% longer. A single individual braking can impact traffic across the city, triggering a slowdown or even complete gridlock. With the help of sensors and cameras powered by cutting-edge software, however, autonomous vehicles brake far less often than their human counterparts and, as a result, are far less likely to cause these traffic disturbances. Even deploying a few autonomous vehicles can have a positive effect on traffic congestion by helping to moderate the speed of the human drivers they share the road with.
Fuel Efficiency and Sustainability
Autonomous vehicles can also improve fuel efficiency over human drivers by controlling their speed and acceleration and by traveling closer together so as to improve air drag and reduce fuel consumption. According to MIT News, if every vehicle on the road were autonomous, not only would travel speeds be boosted by 20%, but we would see fuel consumption reduced by 18% and carbon dioxide emissions lowered by 25%. This development would be pivotal in our continued efforts to bring sustainability to untold numbers of industries and businesses. A study by TuSimple found that their autonomous trucks were 11% more fuel efficient than those piloted by human drivers. This increased fuel economy will allow goods and services to become less costly to consumers while also aiding these companies in their efforts to make their operations greener and more sustainable.
A Look Under The Hood
The level of technological advancement that allows these autonomous vehicle systems to operate have been decades in the making. Arrays of sensors, including cameras, radars, and LiDARs feed data into neural networks designed to mimic the human brain and perform object detection and image segmentation. These neural networks then process this sensory input, including the presence of other vehicles, road signs and obstacles, in order to create a comprehensive surround map of the vehicle’s environment. The next step is then motion planning, where detailed routes and trajectories are calculated using a comprehensive analysis of all the previously collected data. Even then, these processes all still need to account for unseen situations and be able to adapt in real-time to these circumstances. Due to the enormous number of intricate and detailed processes that go into the development of these systems and software, no two are alike and each of these AV systems has their pros and cons.
Forks In The Road
The two primary approaches to the development of autonomous driving are HD maps versus HD map-less systems. The benefit of using maps is in its simplified object detection and motion planning, but these systems are dependent on continuous communication for data updates and are prone to obsolescence. HD map-less systems, like the one developed by autonomous driving software company Imagry, rely almost entirely on real-time data and are more in line with how human drivers operate. They also are more self-sufficient and less vulnerable to cyber threats, but do require advanced onboard perception capabilities and complex real-time processing. After this initial split in philosophy, there exists several others that have been at the forefront of some debate in the industry. Rule-based vs. Neural Network-based motion planning is one such sticking point with safety and regulatory bodies preferring the more definable “if-then” approach that is the hallmark of rule-based systems. While the construction of predefined scenarios offers high explainability, these systems struggle to adapt to new, unforeseen situations, an area where neural network-based systems excel.
The Road Ahead
The groundwork continues to be laid to enable the widespread adoption of autonomous vehicles all over the world. There is certainly no shortage of automakers and companies willing to invest billions of dollars into the development of autonomous vehicles and services centered around them. While there remain many different systems and processes that go into the creation of self-driving vehicles, all experts in some capacity agree on the vast amount of practical benefits that autonomous vehicles and their implementation can have for society. The next and perhaps most important hurdle to clear is building up the general public’s trust in these technologies. The advancement of artificial intelligence also began beneath a cloud of skepticism and mistrust that had to be overcome. Now, there is not a major industry or company in the world that does not utilize these technologies in some capacity or another. Autonomous vehicles will have a similar hill to climb, but as these systems advance and become more prevalent on our roads, our comfortability and familiarity with them will also only increase. As these technologies advance at a rapid pace, the AV industry is farther down the road to global adoption than some might think.
Credit: Source link
