It's a given that autonomous cars are on the way, but the technology's early iterations will likely need a lot of improvement to handle busy environments.
Driverless cars have the potential to solve pressing transportation problems. Computerized, autonomous driving systems can place cars in the same lane in extremely close proximity to reduce congestion, and they can be designed to emit fewer pollutants than traditional vehicles. Wireless vehicle-to-vehicle communication also enables them to avoid one another while en route, preventing collisions.
Multiple countries have either already deployed autonomous vehicle projects or are currently in the process. According to Sky News, Japan, Singapore and Germany have already run test trials, with Sweden soon to follow. The United States hosts multiple driverless car projects, and the United Kingdom will launch its own in 2015. Vendors include Google and a host of commercial automakers, including Audi, Nissan, Toyota and Mercedes-Benz.
Demand for the technology will prompt governments to adapt. In the United States, for example, Michigan, Nevada, Florida and California already have authorized them, and companies and research groups will take advantage of the opportunity. The Los Angeles Times reported that by 2025, as many as 230,000 new autonomous vehicles could hit world roads each year.
Here are three projects that represent a potential shift in the way people travel on land:
Professor Raj Rajkumar, co-director of the General Motors-Carnegie Mellon Autonomous Driving Collaborative Research Lab, and his team began testing a modified Cadillac SRX in Washington, D.C., this summer. They installed cameras and sensors to detect traffic lights and pedestrians, and read road signs. Drivers could take control of the car themselves if they wished, and the steering wheel vibrated to alert them when they had to take manual control. The university transported congressmen in the vehicle in June to demonstrate the car’s effectiveness in hectic local traffic and continued successive testing thereafter.
Last month, the California Department of Motor Vehicles issued 29 permits to companies that allowed them to test autonomous vehicles on state roads. Google won 25, and Audi and Daimler AG-Mercedes Benz won two each. Google's autonomous vehicle project includes cars equipped with cameras that record external visual cues like stop lights, a LIDAR laser that maps vehicle surroundings, and radar sensors that detect nearby objects. This information is sent to Google Chauffeur, a computer that drives the car unless a user presses a red button, turns the steering wheel, or applies the brake pedal to take control manually. Automakers see enormous potential in California for self-driving cars because of the expansive, diverse terrain with countless roads, and the millions of residents who could use the technology.
The town of Milton Keynes announced in 2013 that the government would fund its £1.5 million project (approximately $2.4 million) to test driverless cars in 2015. The Transport Systems Catapult organization is managing the project and will deploy the Milton Keynes vehicles as part of its low-carbon urban transport zone (LUTZ) Pathfinder program. The TSC selected the RDM group to build electric self-driving pods, which will carry up to two passengers on pavement at up to 12 kilometers per hour. The government hopes the cars will reduce congestion and avoid hitting humans, and if the 2015 trials are successful, up to 100 pods will be deployed by 2017.
It's a given that driverless cars are on the way, but the technology's early iterations will likely need a lot of improvement to handle busy environments. Some scientists believe that the current group of vehicles will have difficulty understanding situational context. Several scientists at the 2014 Automated Vehicles Symposium told the MIT Technology Review this summer that they expected self-driving cars to be limited to specific, well-controlled settings like construction sites and special environments with low speed limits and traffic. People, for example, anticipate actions based on other vehicles' behavior and eye contact with pedestrians and other drivers, which would be more difficult for a computer system to make sense of. The conference organizers asked 500 attendees when they'd trust a robotic car to take their children to school, and more than half said that by 2030 at the earliest, one-fifth said 2040, and roughly one in 10 said never.