谷歌無人駕駛汽車發生交通事故

One of Google’s driverless cars has been involved in an accident for the first time in a situation where human drivers rely on the normal social etiquette of the road to avoid scrapes, according to an accident report filed in California.

美國加州一份交通事故報告顯示，谷歌(Google)無人駕駛汽車首次捲入人類駕駛員依據避免刮蹭的常規道路交通禮儀可避免的交通事故。

The incident highlights the challenges of programming machines to deal with the sort of social situations that humans encounter every day on the roads, but which are beyond robots to resolve, experts said.

專家表示，這起事故凸顯出了智能設備在應對各類社會場景方面面臨的挑戰，這些社會場景是人類每天在路上都會遇到的，但它們卻超出了機器人的解決能力範圍。

It also threatens to cast an unflattering light on the technology at a time when Google is fighting a proposed California rule requiring autonomous vehicles to have human drivers on hand to override their controls in emergencies.

這起事故還有可能對谷歌的無人駕駛技術帶來負面影響。眼下，谷歌正在與加州的一項法規草案作鬥爭，該法規要求無人駕駛汽車必須配有人類駕駛員，以便在緊急情況下以手動方式控制汽車。

However, the company suggested that the accident would have occurred even if a human had been at the wheel, while adding that it had already adjusted its algorithms to take account of the lessons from the scrape.

但谷歌暗示，即使人類駕駛員當時接管了汽車的控制權，這起事故依然會發生。該公司還表示，已從這起剮蹭事故中吸取了教訓，並據此調整了其算法。

“There are all sorts of social clues that people abide by that are hard to program into a car,” said Jerry Kaplan, a serial technology entrepreneur and author of a recent book on robotics. “There was nothing terrible about this, no one got hurt or killed. But it will be a long time before the cars can be driven in heavily used streets and abide by the same social conventions people use.”

連續創建多家科技企業的傑裏•卡普蘭(Jerry Kaplan)表示：“人類遵循的一些形形色色的社會線索，很難被編入汽車程序。這曾經沒什麼可怕的，沒人因此受傷或死亡。不過，現在距無人駕駛汽車在繁忙的街道上行駛並遵守人類遵守的社會習俗，還有很長一段時間。”卡普蘭近期出版了一本有關機器人的著作。

In a filing with the California Department of Motor Vehicles, Google said its car had been trying to pull out into slow-moving traffic on Valentine’s Day when a bus driver did not slow to leave room, causing a minor collision.

谷歌在提交給加州機動車輛管理局(Department of Motor Vehicles)的一份文件中表示，情人節這天，當一輛谷歌無人駕駛汽車試圖變道併入緩慢移動的車流時，一名大巴司機並沒有減速留出空間，導致兩輛車發生輕微碰撞。

In a blog post on Monday, it blamed the accident on a “tricky set of circumstances” caused partly by an adjustment it had made to the car’s algorithms weeks earlier to help it drive more like a human.

在週一發表的一篇博文中，谷歌將這起事故歸咎於“一系列棘手的情況”，而這些情況在一定程度上是由幾周前谷歌對無人駕駛汽車算法做的調整導致的，那次調整旨在使其汽車的駕駛方式更像人類駕駛。

“This is a classic example of the negotiation that’s a normal part of driving — we’re all trying to predict each other’s movements,” the company said. “We clearly bear some responsibility.” However, it also said that a Google employee in the car had also believed there was room to pull out.

“這是一個關於協商的典型例子，而這種協商是駕駛的一個正常組成部分——我們都試圖在預判對方的運動。”谷歌稱，“我們顯然負有部分責任。”但它還表示，當時坐在車內的谷歌員工也相信有空間可以併線。

The car had pulled close to the side of the road to make a turn, leaving room for other vehicles that were carrying on straight ahead to pass in the same lane — a manoeuvre Google said it had programmed into the car to make it a more considerate driver.

當時爲了轉彎，這輛汽車已靠近路邊行駛，爲同一車道內仍保持直行的其他車輛讓出空間——谷歌表示，它已將這種機動動作編入汽車程序，以使其成爲一個更顧及他人的駕駛者。

After finding its path was blocked by sandbags that had been placed round a drain, the car was forced to move back out into the line of traffic, causing the accident.

在發現前方道路被一堆沙袋（堆放在一條下水道周圍）阻擋後，這輛谷歌汽車不得不駛回車流當中，由此造成了這起事故。

Google said it had already programmed lessons from the accident into its software, along with its findings from running simulations of thousands of similar situations.

谷歌表示，它已將該事故教訓寫入其軟件中，一併寫入的還有從對數千個類似情形的模擬中得出的研究發現。