2017年3月13日 星期一

Intel Buys Mobileye in $15.3 Billion Bid to Lead Self­-Driving Car Market

Intel Buys Mobileye in $15.3 Billion Bid to Lead Self­-Driving Car Market

By MARK SCOTT MARCH 13, 2017

In the world of driverless cars, household names like Google and Uber have raced ahead of rivals, building test vehicles and starting trials on city streets. But when it comes to what is under the hood, an array of lesser­-known companies will most likely supply the technology required to bring driverless cars to the masses. And in a $15.3 billion deal announced on Monday, Intel moved to corner the market on how much of that technology is developed.

The chip maker’s acquisition of Mobileye, an Israeli company that makes sensors and cameras for driverless vehicles, is one of the largest in the fast-growing sector and sets the stage for increasing competition between Silicon Valley giants as well as traditional automakers over who will dominate the world of autonomous cars. The likes of Google and Uber have already invested billions of dollars in their own technology, signing partnerships with automakers like Chrysler and Volvo and sending test vehicles onto the road in a bid to cement their place in the industry. The sector is estimated to be worth $25 billion annually by 2025, according to Bain & Company, a consulting firm.

Faced with an existential threat to its legacy computer business, Intel — alongside competitors like Qualcomm — has focused on autonomous cars as a new and potentially lucrative market. Many of these driverless vehicles, experts say, will require immense computing power, including the latest microchips able to crunch reams of data in seconds to keep the cars safe, and on the road. And by acquiring Mobileye, whose digital vision technology helps autonomous vehicles safely navigate city streets, Intel aims to broaden its offerings beyond just chips to a wider suite of products that driverless vehicles will require. It hopes, as a result, to appeal to automakers that want to offer autonomous driving but lack the in­-house expertise and do not want to rely on the likes of Google.

“Scale is going to win in this market,” Brian Krzanich, Intel’s chief executive, told investors on Monday. “I don’t believe that every carmaker can invest to do independent development into autonomous cars.” Intel has a history with such strategic moves. It cornered the personal computer market over more than three decades, supplying hundreds of millions of desktop computers with much of their internal architecture, after dominating which microchips were used in the industry. But in recent years, Intel has struggled to find its feet as people’s habits have increasingly turned to the mobile world, where the company’s chips have lost out to rivals. Last year, for instance, the company announced that it was laying off 12,000 people, or 11 percent of its global work force, as demand for personal computers continued to decline. While Intel still earns more than half of its annual revenue from traditional computing chip-­making operations, the company’s sales from its “internet of things” division, a unit that includes its burgeoning auto-making team, grew 15 percent in 2016, to $2.6 billion, according to regulatory filings.

Over the last 18 months, Intel has signed partnership deals with BMW and Delphi Automotive, an auto parts supplier, to expand its presence in the field. It also acquired a 15 percent stake in Here, a digital mapping business owned by a consortium of German automakers, and announced last year that it would invest $250 million in start­ups working on driverless car technologies.

Mobileye, founded in Jerusalem in 1999, has signed deals with several automakers, including Audi, for the use of its vision and camera technology, which uses machine learning and complex neuroscience to help drivers — and increasingly cars themselves — avoid obstacles on the road. It also has longstanding ties with Intel. The chip maker was a partner with Mobileye and BMW last year over efforts to bring autonomous cars to city streets by 2021.

In January, the companies announced plans to have up to 40 autonomous cars on American and European roads by the end of this year as part of initial trials. As that collaboration grew, Intel and Mobileye executives began talking about a potential takeover at the end of December, holding meetings, mostly in New York, to complete a deal in which Mobileye’s executives would take the lead in new expanded efforts. “This deal makes Intel a Tier 1 partner for the automotive industry,” said Martin Birkner, an automotive analyst at Gartner, a technology research company in Munich.

 “As the industry moves toward autonomous driving, new types of digital suppliers like Intel are developing quickly.” Intel’s efforts to stamp a claim on driverless cars represents a recognition that chip-­making rivals like Nvidia and Qualcomm had moved slightly ahead in the race to provide the computing power needed for autonomous vehicles.

As part of the deal, Intel said it would buy Mobileye’s outstanding shares at $63.54 a share, a 34 percent premium to Mobileye’s closing price on Friday. The acquisition requires shareholder and regulatory approval, and is expected to close by the end of this year. Much of Intel’s success will depend on Amnon Shashua, Mobileye’s cofounder and chief technology officer, who has a doctorate in brain and cognitive sciences from the Massachusetts Institute of Technology.

Part of Mr. Shashua’s plan is to have cars with Mobileye’s advanced driver assist systems collect imaging and location data that can be used to create what the company calls Road-Book — a vast digital map of roadways in the United States and Europe. The goal, Mr. Shashua said, is to provide carmakers with a complete product line of digital services that go beyond what they can do for themselves. “The collaboration that we want to do can’t happen if we are two different organizations,” he said on Monday. “The collaboration already runs deep.”

Still, experts say autonomous cars are unlikely to hit the roads by the end of the decade, at the earliest, because regulators are beginning to question which rules such cars should follow and because of struggles to make the technology work seamlessly. Uber, the ride­-booking service, halted its driverless car tests in California after local officials said the company did not have the required permits, though its tests in Pittsburgh are continuing.


Google’s own efforts ran into difficulties after the company’s driverless cars were involved in a spate of collisions. And in Europe, regulators are divided on the issue of self-­driving cars, causing the automotive industry to complain that the delays could hamper plans to take the technology to the streets there. Still, with technology companies and global automakers making hefty — and costly — bets on autonomous cars, experts that say more deals like Intel’s acquisition of Mobileye are likely to follow as firms jostle for position. More acquisitions, said Mr. Birkner of Gartner, “are an absolute necessity.” “Carmakers and Silicon Valley companies,” he said, “are realizing that they both bring different skills to the table.”

2017年3月9日 星期四

發展人工智慧 宜具遠見規畫

發展人工智慧 宜具遠見規畫

2017-03-10

聯合報 趙文衡/台灣經濟研究院副研究員(新北市)


行政院國發會為了瞭解產業趨勢,引進華碩機器人Zenbo,展現對創新產業的重視;國發會認為Zenbo如能再精進,對居家照護產業應是很大進展。

以往自動化威脅製造業勞工,現在服務業也可由機器人取代,勢必衝擊整體就業。政府不能只強調新科技發展,而忽略其對勞工與社會的衝擊。

有專家認為,根據三次工業革命經驗,科技進步只會讓工作機會增加,即使一些工作消失,但仍會創造許多新的工作機會。但如果仔細分析軌跡,可發現正進行的第四次工業革命,創造就業的動力已不復存在。

第一與第二次工業革命,機器雖取代部分勞工,但由於全球化啟動,原本只在本地銷售的產品,可賣到世界各地,市場擴張使生產規模提升,造成對人力需求大增。第三次工業革命伴隨服務業興起,服務業就業大增,彌補製造業喪失的工作。以台灣為例,台灣服務業就業占比由一九八○年的卅八%,增加至二○一五年的五十九%。

第四次工業革命,無法再找到另一個全球市場。同樣的,欲找到跟服務業一樣規模的新產業難上加難。

一些統計調查更可證明,未來新增工作將無法彌補失去的。根據日本三菱綜合研究所指出,二○三○年,以人工智慧應用將取代日本七百四十萬個工作,增加五百萬個新工作;相抵後,就業人口減少二百四十萬。世界經濟論壇的研究也指出,新的工作機會只能彌補三成失去的,其他七成勞工將無工作可做。

以往服務業是就業者的大本營,可是機器人取代的不只是製造業,同時也是服務業,這對就業是嚴重警訊。

人工智慧大量運用,也會顛覆經濟政策邏輯,經濟成長價值需重新思考。在人工智慧運用下,雖然產值快速上升,但就業人口卻快速下降,GDP與就業的正向關係將反轉。往後政府必須在促進GDP與增進就業中擇一。

可預期的,大規模失業將引起社會動亂,甚至戰爭。藉由人工智慧獲利者,僅會是少數人,貧富不均將急速惡化。這是人類自己製造的危機,可能也是本世紀需要面對的最重要議題之一。


此時,政府應多花時間發掘並因應這些新科技下的潛在危機,少一點作秀行為,多一點具有遠見的規畫。

2017年3月8日 星期三

走進Uber看台灣 生計之爭vs.改善生活的願景

走進Uber看台灣 生計之爭vs.改善生活的願景
20170308
文/陳泳丞

知名科技公司Uber為市值650億美元的大公司。(/美聯社)
Uber在世界各地開疆闢土,有些順利、有的困難重重,也引發不小爭議甚至既有運輸體系的抗爭。以目前狀況來說,未來Uber在國內能否合法上路仍在未定之天;不過,我們實際走訪舊金山Uber總部,卻也觀察到一些有趣的點。

Uber運用新科技讓叫車這件事變得簡單、有效率甚至還可以很環保(共乘),對於在台灣習慣看到滿街計程車的我們並不稀奇,因為隨手一招就有車,即使是半夜兩三點也是如此;不過在很多國外的大都會狀況就很不一樣,Uber的創辦人也是因當年在巴黎街頭叫不到車,才萌生創建這種可以手機「一鍵叫車」的新平台念頭。

說到Uber在很多地方遇到的麻煩,其實經常都是挑戰到既有運輸體系的「生計」問題,這些體系都有已行之有年的適用法令或管理規則,但Uber卻不容易找到這樣的東西,所以很難「合法」。

否則,講到管理效率(大數據分析可以找到很好的效率運輸搭配)、安全(有定位系統甚至可以監測司機是否超速)、服務(乘客不爽可以手機按鍵立馬投訴)等,如果可以找到大家都能接受的實施細則,對絕大多數民眾都是一件好事。

不過這種對既有體系的衝撞,往往牽涉到輿論、利益甚至政治等層面的考量與角力,公說公有理、婆說婆有理,但在與多位Uber主管、工程師訪談過程中,筆者卻發現有一點很有趣,那就是願景。

看到這家年輕公司裡頭的年輕人,講到發展技術、新應用時,眼神不由自主散發出的光彩非常明顯,主要原因是他們感覺因自己的努力,可以為這個世界帶來一些改變,改變環境(空氣污染)、改變城市移動(交通)的樣貌,目的是要讓人的「生活」變得更好。

台灣現在氣氛有點低迷,除了一些已大放異彩的科技公司徵起人來氣魄十足外,能夠帶給年輕人希望的公司似乎少了點。不管Uber來不來,台灣應該有更多這類多一點創新跟願景的公司。


(工商時報)

2017年3月6日 星期一

科技交通新連結,不能等

科技交通新連結,不能等
20170307
林建甫專欄

蝶戀花旅行社遊覽車翻車意外引發了國內外媒體的關注,除了駕駛的「過勞」、車輛的「保養維護」問題值得政府、企業及社會省思外,車輛的行駛紀錄、車輛的安全反應,這些與科技的新連結恐也值得社會注意。

因為「智慧運輸」的時代已經來了。UBER的在台停止服務,讓科技界覺得台灣有反智的現象,然而因為稅負及保險的爭議而拒絕UBER是講得過去。可是應用高科技的車聯網、無人車,或更大的智慧運輸系統,台灣必須急起直追,運將也必須面對挑戰。因為許多關鍵技術已逐漸發展成熟,不管你願不願意,在不久的未來,到處都可見智慧運輸系統,馬路上跑的都將會是透過「車聯網」管理的無人車。

今年的世界行動通訊大會(Mobile World CongressMWC2月底在西班牙舉行,過往這是手機產業最大的盛宴,但今年除了捲土重來的Nokia,其他手機並未獲得太多的版面。反倒是「5G應用」搶盡所有鋒頭,特別是車聯網相關的產品應用,整個展場好像化身為「車展秀」,令人耳目一新。

智慧運輸的關鍵在於「溝通」,包括與使用者的溝通,以及車對車(V2V)、車對交通設施(V2I)、車對雲端中心的外部溝通。其實這當中最難的恐怕是與人的溝通。不可諱言,目前車載系統的數位助理服務距離電視《霹靂遊俠》中的「夥計」仍有些遙遠,可是從蘋果推出Siri、安卓的Cortrina到亞馬遜的Alexa,目前有越來越多車廠開始將數位語音助理服務應用在車載系統,讓使用者能透過自然口語方式操作車輛功能。未來的數位語音助理服務可藉由「深度學習」,讓系統更能了解駕駛的實際需求,變得更「聰明」。另外其他車子需要的互聯、檢查、感應,乃至於駕駛等其他的溝通,拜物聯網之賜,都已不是那麼困難。

目前自動駕駛汽車所採用的識別系統,早已不是簡單趨近感應器。先進駕駛輔助系統(Advanced Driver Assistance SystemsADAS)結合了感測元件、攝影鏡頭,以及用於快速、精準測距的雷射光車頭燈設備,不僅能比人眼更清楚地看清前方路況,更能隨時緊盯汽車四周的情況。除此之外,完整的ADAS除了看清車外情況,還能隨時檢查車內狀況,包含整個車輛內部重要運作機件,燃油管線、胎壓、煞車等可以第一時間提出警告。

然而,馬路上有各種突如其來的狀況,在接收到道路上真實情況後,必須有一個大腦能馬上做出準確的安全駕駛回應,這部分必須透過龐大數據資料訓練而成的人工智慧,而這也是無人車是否能安全上路的核心關鍵,吸引了眾多的企業投入研發。雖然有人認為AlphaGo戰勝人類棋王,其實它並不會思考,只是透過機率的計算,選擇最可能贏的下法,但人工智慧在蒐集及儲存訊息、綜合計算等單調反應上,是很容易打敗人類的。

準此,「無人車」不是一個全新的產業,許多的技術不是單一產業能夠完善,必須依賴跨界的合作,例如Google的無人車公司Waymo以平台的方式與車廠合作,百度(Baidu)和英偉達(NVIDIA)宣布合作開發自動駕駛的平台、與福特共同投資雷射雷達公司Velodyne。除了這些大企業大力投資這個產業,其實還有許多新創公司投入這個市場,例如:以色列新創公司Oryx Vision開發的奈米天線感測技術,估值近16億美元的Quanergy專攻研發固態LiDar,都成功地打入無人車的供應鏈中。

綜上,由於無人車涉及眾多領域、技術複雜,背後潛在的商機大到難以想像。目前全球無人車的發展如火如荼,橫跨相當多的產業,當今全球汽車產業1年大約銷售8800多萬輛,成長率僅23%,不過麥肯錫預測接下來汽車產業將進入快速升級的階段,並逐漸向無人車靠攏,預估汽車產業在2030年的市場規模將提升30%,創造高達15兆美元的龐大商機。

最近台灣在車聯網的布局也傳出好消息,台灣大哥大繼去年10月宣布與特斯拉(Tesla)合作建構車聯網平台,在今年的MWC中也公開與全球電信龍頭AT&T合作的好消息,成功切入全球22品牌車廠的加值應用聯盟。未來,我們期望政府能夠從科技來管理運輸業,來帶動服務,同時也能夠更加支持、發展相關的產業。

(作者為台灣經濟研究院院長、台灣大學經濟系教授)


(中國時報)