Tencent Technology News, May 24, seven years ago, Waymo, a subsidiary of Google’s parent company Alphabet, discovered that flowers, soap bubbles, and highway flares would confuse its self-driving cars and brake suddenly. In years of testing, research has found that self-driving cars are becoming more and more distracting. Although their ability to drive automatically on the road has improved, they are still difficult to compare with the capabilities of human drivers. Facts have proved that the chaotic road traffic in the United States is still a daunting place for self-driving cars. The Wizards of Silicon Valley once predicted that by 2021, people should be able to commute to work in self-driving cars. However, the reality is that court disputes, casualties, and tens of billions of dollars in investment have all been spent on a frustrating and capricious technology. Many researchers said that it may take 30 years for autonomous driving technology to become the “next big thing” in the industry. Now, the pursuit of self-driving cars is undergoing a restart. Ride-hailing giants Uber and Lyft are worried about spending too much money in the pursuit of autonomous driving technology, and they are already somewhat powerless. Only companies like Waymo with strong financial support, auto industry giants and a few start-up companies can continue to compete for hegemony in this game. At the end of April, Lyft sold its self-driving car division to Woven Planet, a Toyota subsidiary, for a transaction value of US$550 million. Last December, Uber sold its self-driving car division to another competitor. In the past year, three well-known autonomous driving startups have sold themselves to companies with adequate budgets. Tech giants and auto giants may still continue to develop their self-driving car projects for many years. According to estimates by Pitchbook, a research company that tracks financial activities, each company will spend an additional US$6 billion to US$10 billion before the adoption of autonomous driving technology. But even such a forecast may be too optimistic. Google’s self-driving car project was spun off to form Alphabet subsidiary Waymo, an early engineer Chris Urmson (Chris Urmson) said: “This is a change that will take place in 30 years or more.” He is now Is the CEO of Aurora, the company that acquired Uber’s self-driving car division. So what went wrong? Many researchers will not say anything, this is how science works. You cannot fully predict what will happen in the experiment. The self-driving car project happens to be one of the most hyped technical experiments in this century. Many of the most well-known companies conduct road tests on highways across the United States. This hype has attracted billions of dollars in investment, but it has also shaped unrealistic expectations. In 2015, Elon Musk, CEO of electric vehicle manufacturer Tesla, said that only two years were left before fully functional self-driving cars were born. However, more than five years later, Tesla only provided its cars with a simplified version of self-driving software designed for highway driving. Nevertheless, after several fatal crashes (which the company blames on the abuse of this technology), this has also caused a lot of controversy. Perhaps no company has experienced the turbulence of driverless car development more than Uber. The company hired 40 robotics experts from Carnegie Mellon University and spent US$680 million to acquire a self-driving truck startup, but then it was in trouble. Waymo accused former executives of stealing trade secrets and handing them over to Uber, and the two companies finally reached a settlement. A pedestrian in Arizona was killed by an Uber self-driving car. In the end, Uber basically dumped its autonomous driving department to Aurora at a loss. But for companies with deep pockets, they want science to continue to make rapid progress. In October last year, Waymo achieved a key milestone, namely the launch of the world’s first “self-driving” taxi service. In the suburbs of Phoenix, Arizona, anyone can now take a taxi without a driver. But this does not mean that the company will immediately deploy its technology in other parts of the United States. Following the departure of John Krafcik, a veteran in the auto industry, Dmitri Dolgov, who recently took over as Co-CEO of Waymo, said the company considers its Arizona service to be a Important test. He said that based on what he learned in Arizona, Waymo is building a new version of autonomous driving technology that will eventually be deployed in other areas and other types of vehicles, including long-distance trucks. The suburbs of Phoenix are particularly suitable for self-driving car testing, where the streets are wide, pedestrians are sparse, and there is almost no rain or snow. Waymo provides remote technical support and roadside assistance support for its self-driving cars. These technicians can help the car get out of trouble through the Internet or arrive at the scene in person. Elliot Katz said: “Autonomous vehicles can be deployed in certain situations today, but they still need to be supervised.” Katz is an attorney who has consulted for many large self-driving car companies Services, and later founded a start-up company called Phantom Auto. When self-driving cars are in trouble, the company provides remote assistance and software for operating self-driving cars. At the same time, autonomous driving technology is not flexible enough to reliably handle the various situations that human drivers encounter every day. They can usually cope with the situation in the suburbs of Phoenix, but cannot cope with the complex traffic environment of the Lincoln Tunnel in New York or the downhill ramp of the 101 highway in Los Angeles. Waymo software engineer Nathaniel Fairfield, who has been involved in the project since 2009, described the interference faced by self-driving cars, said: “You have to peel off every layer to see what the technology faces. The next level of challenge. Your car must perform quite well before it can truly enter the position where it can handle the next most challenging environment.” Like Waymo, Aurora is now also developing self-driving trucks and passenger cars. Although no company has deployed self-driving trucks without safety driver supervision, Urmson and others believe that self-driving trucks will be launched on the market faster than any passenger car designed to transport ordinary consumers. Long-distance truck transportation does not involve sudden braking problems that may plague passengers, and the route is simpler. Urmson said that once you are familiar with a certain section of the highway, it is easier to become familiar with other road conditions. However, it is extremely difficult to maintain autonomous driving even on a long, relatively straight highway, and automatic delivery of food in a relatively small community is even more challenging. Another early engineer on the Google team, Dave Ferguson, said: “This is one of the biggest technical challenges facing our generation.” He is now the president of Nuro, a company focused on delivering groceries. , Pizza and other goods. Ferguson said that many people believe that autonomous driving technology will be improved like Internet services or smartphone applications, but the challenges of robotics are obviously much greater. He said: “If you look at almost every industry that tries to solve difficult technical challenges, you will find that those who tend to participate are a bit crazy and very optimistic. You need to keep that optimistic mentality to get up every day. And try to solve problems that have never been solved.” Uber and Lyft have not completely abandoned their ambitions to develop self-driving cars. Although this may not be profitable for a long time, they still hope to deploy self-driving cars by cooperating with companies that are still developing this technology. Lyft now says that autonomous driving may arrive in 2023. Lyft executive Jody Kelman (Jody Kelman) said: “These cars will be able to travel at a certain speed on limited streets and in limited weather conditions. We will be able to deploy these vehicles very safely, but They may not be able to travel too many places initially.” (Tencent Technology Review/Jinlu)
You must log in to post a comment.