Understand we are in pursuit of self-driving cars may be confused as to understand we are in the pursuit of AI. In the past few years, a flood of companies entering the space and constant news updates, it seems as if a fully Autonomous vehicle is just almost not there. Over the past few weeks have been no different: Uber announced a new CEO and $ 1 billion of investment, is it an automatic driving unit, the Waymo start a cab application to open its services to more riders in Phoenix and the La introduced a new definition of AI chip that promises to unlock the full autonomy.
But an unmanned vehicle to remain in the test version, the car manufacturer has a wide range of different estimates of how many years, we still have to go. In early April, Ford CEO Jim Hackett represents the conservative position, admitted that the company initially”too high to estimate the arrival of Autonomous vehicles.” It also plans to launch its first self-driving fleet in 2021, but have significantly dialed back the capacity. In contrast, Tesla’s chief, Elon Musk, bullishly claiming that self-driving technology will probably be relatively safe human intervention in the car by 2020. “I would be shocked, if not next year at the latest,”he said.
I’m not in the business of prediction. But recently I sat down with Amnon Shashua, the Chief Executive Officer of the case, understand the various challenges to reach full autonomy. The acquisition of Intel in 2017, the Israeli infrastructure of automated driving technology partnerships of more than twenty car manufacturers and became a leading space.
Shashua challenges in technology, regulation and business.
Building a safe car. From a technical point of view, Shashua split of drone technology into two parts: its views and its decision-making capacity. The first challenge, he said, is to build an autopilot system that can Perceived When than the best human drivers. In the US, the current car, the fatality rate is about one death per 1 million hours of driving. No drunk driving or texting, the speed is likely to reduce 10 times. Effectively, this means that the self-driving car, the perceptual system should fail, at an absolute maximum, once in every 10 million hours of driving.
But currently the best driving assistance system mistakenly believes that something in their environment once every tens of thousands of hours, Shashua said. “We are talking about three orders of magnitude gap.” In addition to improving computer horizons, he saw two other necessary components to bridge this gap. The first is to create a redundant sensing system using camera, radar and laser radar. The second is to build a very detailed map of the environment, making it easier for cars to handle their environment.
Building a useful car. The second challenge is to build a system, the system can make a reasonable decision, how fast push and when alternative know. But defining what constitutes a”reasonable”little technical challenge than a regulatory one, says Shashua. Any time a driverless car decision, it makes the trade between security and usefulness. “I can be completely safe if I’m not driving or if I drive very slow,”he said,”but then I’m of no use, and society does not want these vehicles on the road.” Regulators, therefore, must formally determine the boundaries of reasonable decision-making, so that car manufacturers can plan their car’s behavior only in these boundaries. It also created a legal framework for assessing blame when a driverless car gets into an accident: if the decision system does not, in fact cannot stay in these boundaries, then it will bear the responsibility.
Build a cheap car. The final challenge is to create a cost-effective vehicle, so consumers are willing to switch to the drone. In the near future, and the technology is still in the tens of thousands of dollars, only one called the car business will be financially sustainable. In this regard,”you are to remove the driver from the equation, the driver costs more than a hundred thousand dollars,”Shashua explained. However, individual consumers may not pay premiums more than a few thousand dollars of technology. In the long term, this means that if the car manufacturer intends to sell driverless cars, they need to find out how to create a more accurate system than exists today for a fraction of the cost. “So the robot-taxi—we are talking about 2021, 2022 time frame,”he said. “The passenger cars will come after a few years.”
Of the case are now working to overcome these challenges in all aspects. It has revised its view of the system, establish a detailed road map, and with regulatory agencies in China, the United States, Europe, Israel, the standardization of rules for self-driving behavior. (And it’s certainly not alone: Tesla, Uber and Waymo are all engaged in a similar strategy.) The company plans to launch a driverless robotic taxi service with the public in Tel Aviv 2022.
This story originally appeared in our wells nomination of AI communications algorithms. It is sent directly to your email Inbox, sign up here for free.