Uber’s driverless future
Car, take me home
To uber somewhere has become a near common occurrence around the world, you simply tap the app and select a driver to come pick you up. Uber CEO Travis Kalanick’s intention to remove a feature of that process could have widespread implications. Mr. Kalanick plans to remove the driver, beginning in Pittsburgh this August Uber will pair users with prototype driverless cars. The service is the beginning of Uber testing the commercial use of driverless technology, using adapted Volvo XC90 cars that have been in development since earlier this year. For now, the cars will still legally require a driver behind the wheel and also have a co-pilot noting the car’s process, but the achievement still stands, as it is the car in control, not the safety driver.
This development has happened quickly by any standards, as it was only in late 2014 that Mr. Kalanick began recruitment for driverless car development. In 2015 he founded the Uber Advanced Technologies Center (ACT) in Pittsburgh which produced a prototype within months. Mr. Kalanick’s success is remarkably fast when compared to the other giants in the field, namely Google and Tesla. Google has been in development for years and Tesla’s autopilot, despite being in commercial use, is still described as being under beta phase testing. Uber’s speed is likely due to its adaptation of existing cars rather than the development of new cars as others in the field have done. The first prototypes were hybrid Ford Fusions equipped with a variety of data collection equipment were used to collect road and map information. Now, Mr. Kalanick has signed an agreement with Volvo establishing a $300 million project to develop driverless vehicles. Volvo has already delivered the first of these, adapted XC90’s, with a hundred promised by the end of 2016. This will make Uber the first company to bring a driverless fleet to the market, whether it will be successful remains to seen.
Tests by the prototypes do seem promising. Over the past year they have been collecting extensive data about Pittsburgh, not just road location, but even potholes and fire hydrants. This allows the cars to effectively navigate the streets, with detection systems and other safety features in place to prevent collisions. Pittsburgh is also an interesting test location, as it is a particularly challenging city to navigate. Unlike other large American cities such as New York which operate on grid systems, Pittsburgh has a hilly layout, cut up by rivers, which mean it has an assortment of winding dead ends and bridges that are challenging to driverless cars. Its success in Pittsburgh would mean it could easily be successful in the large grid system cities where Uber is most used.
Pittsburgh, a city with a longstanding robotic development culture, is also especially welcoming of driverless cars. The autonomous car industry faces difficulty with the lack of laws of or vague laws that can be applied to driverless cars. The field has grown so quickly that lawmakers have been hard pressed to produce effective legislation to ensure their safe usage in public. Uber’s success in Pittsburgh could be an important step in getting driverless cars legalized country-wide, an as yet undealt with step in the autonomous car market. It is important to consider their legality before there are conflicts between large companies attempting to release their fleets and states attempting to prevent them due to uncertainty. If companies are blocked for too long, funding for autonomous cars may falter and stop the steady progress that the market has made so far.
This should be the last thing to happen. Mr. Kalanick claims that around a million people die each year in car accidents for which the driver was responsible. Driverless cars would be much safer. They would also be far cheaper to use, to the extent that it would be less costly to simply uber everywhere than it would be to own a car. Not only would this benefit transport consumers, but it would also likely lead to emissions reductions through reducing the number of cars on the road.
The benefits of driverless cars are clear, but the results are less so. Google’s cars have had road safety issues in failing to account for other drivers, and have been in development without commercial release since 2005. Tesla’s autopilot has proved effective when used as intended, but has had failed through its limitations, as accidents frequently occur when used improperly since human drivers believe to be far more extensive automation that it actually is. Yet, Mr. Kalanick’s efforts seem promising, not only due to the prototype success so far but also given the talent he has been able to recruit to his efforts. He is set to acquire Otto, an autonomous truck company, and with it will gain many of the original researchers at Google who left due to frustration with the slow process. Among them is Anthony Levandowski, who will head Uber ACT. Together with Otto and Volvo, Uber appears set to begin producing driverless cars adapted from existing models.
Mr. Kalanick can therefore confidently state his aim to release a road-ready car by 2021, along with Volvo. This is a claim echoed by other large carmakers with experience in car production. Ford and BMW also both promise fully autonomous cars by 2021. Although previous promises have fallen short, this is the first time it is major car manufacturers and not just Silicon Valley start-ups that make this claim. With increased investment in the field, such claims might prove true. Uber’s experiment is not simply a test of a commercial model, but rather it could show what the field could possibly achieve, and success could help pave the way for legislation of driverless vehicles by 2021.
Further note – distrust of AI simply because it is AI
Some field experts in the area of car safety, as well as legislators, are calling for incredibly strict testing and road safety measures for autonomous vehicles, much more stringent than those imposed on standard vehicles. Some, such as insurance and other legal issues are genuine concerns but they are more associated with legal difficulty rather than safety. There are many safety reasons given for such stringency, but suffice to say most of them are wrong. The belief that we can trust human drivers, who are far more prone to error account for around 70% of car accidents, a million a year, but not autonomous drivers simply because they aren’t present in a tangible sense is absurd.
Andre Hirschler 