福特自動駕駛汽車有限責(zé)任公司(Ford Autonomous Vehicles LLC)CEO Sherif Marakby表示,福特自動駕駛團(tuán)隊將像當(dāng)年福特T車型顛覆世界一樣顛覆自動駕駛行業(yè)。

1907年,芝加哥制襪有限公司在底特律的科克鎮(zhèn)(Corktown)街區(qū)新建了一座分廠。(不到一年后,亨利·福特推出了劃時代的T型車。) 今天,除了穿梭在前方人行橫道上的Spin電動滑板車以外,這座位于密歇根大街上的三層磚房從外面看似乎一如往昔,但事實上,里面早已脫胎換骨,如今這里成了一片創(chuàng)新的熱土。如果亨利·福特生活在21世紀(jì),他一定會對這里一見傾心。
開放式格局,沒有辦公室,可移動的辦公桌,幾間狹小的會議室,隨處可見的白板,嘈雜的環(huán)境,這就是福特為自動駕駛而專門新設(shè)的業(yè)務(wù)部門——福特自動駕駛汽車有限責(zé)任公司。2018年5月開業(yè),這個被它的220名員工戲稱為“The Factory(工廠)”的新部門計劃于2021年推出福特的第一款插電式混動自動駕駛量產(chǎn)車型。同樣駐扎在這里的還有福特的電動汽車先進(jìn)研發(fā)團(tuán)隊——愛迪生團(tuán)隊。目前,愛迪生團(tuán)隊正和福特自動駕駛汽車公司一起在推進(jìn)福特數(shù)百億美元的網(wǎng)聯(lián)多模式出行項目。鑒于城市生態(tài)系統(tǒng)將是電動汽車和自動駕駛首先應(yīng)用最多的地方,福特的高管希望這兩個團(tuán)隊能在市區(qū)辦公,或者更理想的,能在市區(qū)生活。
對此,福特自動駕駛汽車公司的CEO Sherif Marakby表示贊同,他熱情地說道,“把公司建在這里非常合理,我們需要靠近客戶。當(dāng)我拿到這個機(jī)會的時候,我?guī)缀跏呛敛华q豫地說,‘我要去!我想待在這座城市里!’我在底特律已經(jīng)生活了28年,但是幾乎沒在這里工作過。”
Marakby的公司肩負(fù)著推動技術(shù)商業(yè)化的重?fù)?dān),這也是他們的熱情所在。Marakby告訴SAE《自動駕駛車輛工程》雜志:“我們希望推出自動駕駛共享汽車。”他的這句話其實也指明了福特對自動駕駛汽車客戶的定位——目標(biāo)客戶將不再是駕駛員。Marakby表示,“福特將致力于提供定制化的自動駕駛客貨運解決方案。我們的全方位自動駕駛汽車服務(wù)將會以app為載體。乘客通過app叫車后,車輛將自行前去載客,并讓乘客知道這是他的車。”
除了推動這些變革以外,自動駕駛技術(shù)還有諸多細(xì)節(jié)亟待完善,比如確保車輛在所有環(huán)境下都能遵循路規(guī)、安全行駛。“這就是我們的團(tuán)隊在這個不同尋常的地方工作的主要原因,在這里我們能開啟不同的思路,” Marakby說道。
對于自動駕駛變革所帶來的技術(shù)挑戰(zhàn)和經(jīng)濟(jì)壓力,Marakby也有清醒的認(rèn)識。面對收緊的全球汽車市場,福特和一家名為Argo AI的人工智能公司達(dá)成了一份5年10億美元的合作協(xié)議,Argo AI 的算法和傳感器融合技術(shù)已經(jīng)成為了福特自動駕駛團(tuán)隊的核心資產(chǎn)。而就在《自動駕駛汽車工程》雜志付印之際,有消息稱大眾正在和Argo AI就共同成立一家合資企業(yè)的事宜進(jìn)行洽談,未來大眾亦有可能注資Argo AI。為了進(jìn)一步加強(qiáng)自動駕駛團(tuán)隊的研發(fā)能力,除Argo AI之外,福特還收購了其它一些科技初創(chuàng)企業(yè),比如從事云計算軟件開發(fā)的Autonomic和運輸服務(wù)軟件開發(fā)商TransLoc。
Marakby強(qiáng)調(diào),“說到底,公司還是必須產(chǎn)生經(jīng)濟(jì)效益。我們有我們的利潤表。我把技術(shù)規(guī)格、客戶體驗、生態(tài)系統(tǒng)定為我們的收益指標(biāo)。我也在帶領(lǐng)負(fù)責(zé)整車開發(fā)的項目團(tuán)隊。我們公司涵蓋了從前沿自動駕駛技術(shù)到最終運營及商業(yè)化的全產(chǎn)業(yè)鏈。”
快速推進(jìn)
來自埃及開羅的Marakby曾在馬里蘭大學(xué)攻讀電氣和電子工程學(xué)位,在此期間他學(xué)習(xí)了神經(jīng)網(wǎng)絡(luò)。“1990年的時候,神經(jīng)網(wǎng)絡(luò)還停留在理論層面,學(xué)界研究的是大腦是如何運作的、邏輯是如何傳遞的,以及不同的大腦區(qū)域是如何相互連接的。我記得當(dāng)時上課的時候我在想,‘我理解——我知道如何設(shè)計神經(jīng)回路!’但我不知道怎么應(yīng)用。”在當(dāng)時,神經(jīng)網(wǎng)絡(luò)的應(yīng)用領(lǐng)域還是星星之火。Marakby原先計劃畢業(yè)后搬到硅谷,但是后來他去了福特,擔(dān)任車載娛樂、電動汽車和駕駛輔助項目的工程師。
“每次我想繼續(xù)深造做些不同的事情的時候,福特都會給我機(jī)會,”Marakby說道。在之前SAE舉辦的“混動和電動汽車技術(shù)論壇”上,Marakby介紹了福特的電氣化進(jìn)程,令在場嘉賓印象深刻。他的團(tuán)隊在很短的時間內(nèi)就在迪爾伯恩推出了一款電動車型和四款混動車型,這也為他之后暫離福特,加入快節(jié)奏的Uber奠定了基礎(chǔ)。2016年,Marakby加入Uber,并在那里工作了將近一年時間。
“在Uber工作的時候,我必須保持開放的思維。Uber的工作環(huán)境非常輕松,但是節(jié)奏很快,項目都是迅速推進(jìn),而我也將這種風(fēng)格帶到了我的福特團(tuán)隊。你可能不相信,但是這里也有很多原先在Uber工作的專家,他們的思維方式和工作方式都很不同。”
在福特自動駕駛汽車公司的辦公區(qū),Marakby接受了《自動駕駛汽車工程》雜志的深入采訪。在采訪中,Marakby也介紹了團(tuán)隊在成立后的10個月內(nèi)所取得的成果,其中的重點如下:
2021年福特將推出混動自動駕駛車型,而非電動:“雖然我們正在朝著純電動自動駕駛汽車前進(jìn),但我們還沒有成功。福特即將推出的自動駕駛汽車此前在邁阿密等城市完成了數(shù)千英里的路測,我們通過實地載人載貨訓(xùn)練系統(tǒng)。對于自動駕駛而言,混動車型具有良好的燃油經(jīng)濟(jì)性、續(xù)航里程長、載重大,可以同時搭載自動駕駛汽車所需的電力或其它動力系統(tǒng)。此外,混動車的續(xù)航里程也不會因為溫度變化而衰減。”
“對于交通服務(wù)而言,最重要的就是運行時間,這和賣私家車截然不同。如果不充電,即使是安裝有超大容易電池的電動汽車也無法連續(xù)行駛15到20小時。你不能讓一輛自動駕駛汽車下線8小時去充電,這對于運營商來說是不切實際的。”
自動駕駛汽車對標(biāo):“我們對標(biāo)的是扎根于使用和客戶體驗的航空公司。航空公司深知‘把時間還給消費者’意味著什么,他們知道不同的消費者會有不同的體驗。有的乘客會在飛機(jī)上工作,有的會睡覺,有的打游戲。這點和自動駕駛汽車很相似。有的乘客可能不想使用車輛上的屏幕,有人可能只想用最大的,其他人可能想用自己的設(shè)備。自動駕駛汽車的設(shè)計將不再圍繞著駕駛員,內(nèi)部空間會很靈活。”
福特是否會在2022年采用蜂窩V2X技術(shù):“現(xiàn)在V2X (車對萬物互聯(lián))已經(jīng)吸引了大量企業(yè),福特也是其中一員。我們正在研發(fā)一種‘重載計算’的行車電腦,它將融合所有車載傳感器硬件,而V2X將大大減輕行車電腦的負(fù)擔(dān)。V2X也會為非自動駕駛汽車帶來諸多好處。我們認(rèn)為和城市基礎(chǔ)設(shè)施相連是一個很好的策略,而蜂窩通信將是信息傳播的重要途徑。”
新時代工程師:“新時代的工程師需要擁有不同的技能,而且愿意接受改變。我們的團(tuán)隊成員來自各式各樣的背景,經(jīng)驗也各不相同,有的是剛畢業(yè)的學(xué)生,有的則是來自其它行業(yè)的汽車界新兵。我們正在嘗試結(jié)合軟件開發(fā)的敏捷性和汽車研發(fā)的魯棒性,這一點非常重要。但究竟如何結(jié)合,我們必須審慎考量。我們不想讓新人為了合群而收斂棱角,我們希望他們能帶來獨到的見解。”
“這也是我在這里學(xué)到的。如果你招攬了其它行業(yè)的人才,你必須確保他的身邊有一名了解汽車行業(yè)的前輩,可以幫助新人適應(yīng)新公司,助他獲得成功,這樣也能融合行業(yè)內(nèi)外的專業(yè)知識。”
關(guān)于福特是否會建立自己的操作系統(tǒng):“我們這里有一支軟件工程師團(tuán)隊正在編寫自動駕駛服務(wù)的,主要針對哦自動駕駛汽車服務(wù)提供更好條件、調(diào)度和路徑規(guī)劃。我們將這個操作系統(tǒng)命名為車隊的“orchestration(管弦樂隊)”,它對于自動駕駛汽車而言非常重要,特別是當(dāng)遇到施工區(qū)域、繞路或封路的時候。其中包括了應(yīng)用程序、調(diào)度尋徑,甚至支付功能。我們也是業(yè)內(nèi)首家在開發(fā)自動駕駛綜合操作系統(tǒng)的公司。我們將融合整個生態(tài)系統(tǒng),確保操作系統(tǒng)在任意一個城市的街道上都能安全運行,這樣的能力非常重要。”
關(guān)于邁阿密測試的收獲:“我們發(fā)現(xiàn)我們可以覆蓋一個100平方英里的區(qū)域,在這個區(qū)域里可以做很多事。我們和Postmates、沃爾瑪?shù)绕髽I(yè)合作的測試車準(zhǔn)確地完成了取貨,但搞錯了把送貨的地點!現(xiàn)在我們可以在地圖上看到貨物的位置。不過真正的自動駕駛物流服務(wù)要到數(shù)年后才能落地。邁阿密的測試還在繼續(xù),那里的地圖和剛開始測試時已經(jīng)截然不同。一開始地圖只覆蓋了‘市中心’。我們現(xiàn)在也意識到我們必須了解各家企業(yè)都在運輸哪些物資。”
關(guān)于2019年的工作重心:“2019年是我們推動邁阿密路測中所展示的那些技術(shù)的量產(chǎn)和商業(yè)化的重要一年。此外我們還有大量關(guān)于生態(tài)系統(tǒng)、車輛尋徑、整車開發(fā)、冗余設(shè)計的工作。我們已經(jīng)制定了一個全年執(zhí)行計劃。2019年,我們將大力提高駕駛安全性,建立全生態(tài)系統(tǒng),并和沃爾瑪?shù)绕髽I(yè)合作。聚焦現(xiàn)在,展望未來。從‘我們知道怎么做’過渡到實際應(yīng)用。今年將是我們量產(chǎn)計劃的第一階段。”
關(guān)于自動駕駛汽車的熱潮:“我們知道大家在使用什么硬件,但其實硬件還待改善。自動駕駛技術(shù)還需要借助可靠的商業(yè)模型來提供服務(wù),包括客服中心、應(yīng)用軟件、調(diào)度算法等等,這樣才能確保自動駕駛車輛的安全性,例如遵循路規(guī),實現(xiàn)安全泊車。但現(xiàn)在的自動駕駛車輛還做不到。”
Ford Autonomous Vehicles LLC aims to do for AVs what the Model T did for just about everyone. CEO Sherif Marakby explains.
Henry Ford was not quite a year away from rolling out his world-changing Model T when, in 1907, the Chicago Hosiery Co. built a factory branch in Detroit’s Corktown district. Today the exterior of that same three-story brick building on Michigan Ave. appears unchanged from its sock-sewing past, aside from the Spin electric scooters on the front sidewalk. But inside, it’s abuzz with the innovative stu? that a 21st-century Henry would love.
Open ?oor plan. No offices. Rolling desks. A few tiny meeting rooms. White boards galore. And fairly high levels of ambient noise. Since opening in May 2018, “The Factory,” as its 220 employees call it, has been home to Ford Autonomous Vehicles LLC—the dedicated business unit that aims to launch Ford’s first self-driving, hybrid-electric production AV in 2021. Also based there is Team Edison, Ford’s EV advanced-planning group. Together, they are the vanguard of Ford’s multi-billion-dollar thrust into connected, multi-modal mobility. Ford leadership wanted both teams to work—and ideally, live—within the urban ecosystem where their products initially will be used most.
Sherif Marakby agrees. “It makes so much sense for us to be here, close to the customer,” Marakby, the CEO of Ford Autonomous Vehicles, enthused. “When this opportunity came up, it took me all of about a second to say, ‘I’ll go! I want to be in the city!’ I’ve lived in the area for 28 years but only spent a little bit of it working in Detroit.”
His organization’s role and passion is technology commercialization—”to make autonomy work for the masses,” Marakby told SAE’s Autonomous Vehicle Engineering. It involves defining who the customer is—and that’s not someone driving the car anymore. “With AVs, we tailor the vehicle experience to the passenger and to goods,” he explained. “Providing a full service with AVs involves the app, how you approach the vehicle as a rider and how the vehicle comes to you—and lets you know it’s your car.”
Such changes, along with little matters such as ensuring the vehicle obeys traffic laws and operates safely in all environments, “are a major reason that our teams are located in a di?erent place—so we can think di?erently,” he said.
Marakby’s a realist about the technical challenges and financial burden that come with the changes. In a tightening global vehicle market, Ford is investing $1 billion over five years in Argo AI, whose algorithms and sensor-fusion expertise are vital assets for the Autonomous Vehicles team. (Talks with Volkswagen toward a potential joint venture on self-driving vehicles, including VW investment in Argo AI, were reportedly progressing when AVE went to press.) Acquisitions by Ford Mobility of other tech start-ups, including Autonomic (cloud computing software) and TransLoc (transit-services software) bolster the team’s toolset.
“At the end of the day, my organization has to be a viable business,” he asserted. “We have our own P&L. I set the specification, and the customer experience, and work on the ecosystem. I also direct the program team that directs the vehicle development. Everything that is needed for autonomy to eventually run a service and make money is in this organization.”
Moving fast
A native of Cairo, Egypt, Marakby studied neural networks at the University of Maryland, as part of his degrees in electrical and electronics engineering. “At the time, in 1990, neural nets were theoretical—how the brain works, how logic ?ows and how you can connect the parts. I remember taking classes thinking, ‘I can grasp this—I know how to design circuits!’ But I didn’t know how to apply it.” Few did then. He’d planned to move to Silicon Valley after graduation, but landed at Ford, building his engineering career in infotainment, electrified-vehicle and driver-assistance programs.
“Every time I wanted to expand and do something di?erent, Ford gave me the opportunity to do it,” Marakby noted. Attendees of SAE’s early Hybrid & EV Symposia will remember him presenting Ford’s electrification progress. Launching an EV and four hybrids within a short timeframe in Dearborn somewhat prepared him for the rapid work pace at Uber, where Markaby detoured for about a year in 2016.
“I had to open my mind when I was there,” he said. “It’s a very casual but super-fast environment. It moves fast to get the job done. And that wasn’t hard to translate into Ford, believe it or not. It’s [Uber] a group of people, like those who work in this building, who think and work di?erently.”
In a wide-ranging interview at The Factory, Marakby discussed progress his team has made in its first 10 months and challenges ahead. Some highlights:
On a hybrid AV, rather than electric in 2021: “We’re on a technology journey toward the battery-electric AV, but we’re not there yet. We modeled the upcoming AV on learnings from our thousands of miles of carrying people and goods in Miami and other cities. Hybrid propulsion o?ers great fuel economy, range, and the ability to carry the significant loads— electrical and otherwise—of an AV. Hybrids can work in all thermal conditions without losing range.
“In transport services, it’s all about uptime, which is very di?erent than selling a vehicle. Even a huge battery in an EV will not be able to run 15 to 20 hours a day with a 50 percent loss due to loads without charging. You can’t take an AV offline for eight hours to charge it; that’s not practical for operators.”
On AV benchmarking: “We benchmarked the airline industry, which is built on utilization and customer experience. They know what giving time back to the consumer means, and that the experience is di?erent between customers. It may be more time for working. For sleeping. For playing games. The customer may not want to use any of the screens in the vehicle; or only the big one; or they may want to use their own device. By not having everything designed around the driver, the interior space becomes ?exible.”
On Ford adopting cellular V2X in 2022: “There is already critical mass around V2X (vehicle-to-everything connectivity) and we believe in it. We’re developing a ‘heavy compute’—all the sensor hardware on board—and V2X will lighten up that on-board compute tremendously. For non-AVs there are a lot of benefits to V2X as well. We feel that working with cities and infrastructure is a good path for us and through cellular communication is how we’ll be able to give information.”
New-age engineers: “You need a mix of skills and a willingness to change. It’s combining background and experience with new talent that tends to be right out of school or comes from a di?erent industry. Mixing the two is really important—we’re trying to bring the agile-software and robust-automotive mindsets together. But you must be careful how you mix. You’re not asking people to blend in; you’re asking them to bring their thinking to work with other people.
“I learned that in this environment, when you bring in someone from outside the industry, make sure they have a buddy, someone from in the team that understands the industry so they can navigate their way around the company. Mix the internal and external knowledge while making the new folks successful.”
On Ford deciding to “own” the O/S: “We have a group of software engineers here who are creating the code for, lack of better terms, dispatch and routing of the AV service. The operating system. We call it ‘orchestration’ of the ?eet and it’s important for AVs, particularly when they encounter construction zones, detours and closed roads. We do the app, the dispatch routing and even payments. Nobody’s done this before for AVs. Owning this capability is very important—integrating the whole ecosystem and making sure it works every time is inherent to working with AVs on the streets of any city.”
On learnings from the Miami testing: “We learned we can cover a 100-square-mile area and do a lot of work. Partnering with businesses such as Postmates and Walmart, we got the pick-ups pretty right…and the deliveries pretty wrong! But now we can map the streets showing where the deliveries are, years before we begin actual service. The map of Miami where we’re operating now is very di?erent from where we started. It’s no longer ‘downtown only.’ Now, we have to understand each business, what they’re delivering.”
On 2019’s focus: “It’s a big year for industrializing and commercializing what we showed in Miami. A lot of activity around the ecosystem, routing the vehicle, vehicle development and engineering all the redundancies. We have execution plans throughout the year; 2019 gets ready, with safety drivers, to build the whole ecosystem, run the businesses with Walmart and other companies. Zoom in on where and how we’re going to operate. Take the step from ‘we know how to do it’ to actually doing it. The first phase of building it to scale is this year.”
On AV hype: “We know what everybody is using for hardware—and it’s just not ready. And all of this tech needs a viable business to provide the services—with call centers, apps, dispatch algorithms that make sure the vehicle obeys all traffic laws, parks itself safely, etc. AVs can’t do that yet.”
Author: Lindsay Brooke
Source: SAE Autonomous Vehicle Engineering Magazine