德爾福汽車公司的業(yè)務(wù)服務(wù)部門位于美國(guó)加里福尼亞州的山景城。8月1日,該部門副總裁Glen DeVos宣布,將在新加坡開展一項(xiàng)廣泛的自動(dòng)駕駛汽車試點(diǎn)項(xiàng)目,旨在2019年底前實(shí)現(xiàn)基于云技術(shù)的“SAE 第4級(jí)-自動(dòng)駕駛”。
該項(xiàng)目預(yù)計(jì)將于2022年正式開始。目前,德爾福正在與新加坡道路交通運(yùn)輸管理局(LTA)合作進(jìn)行一項(xiàng)分階段完成的項(xiàng)目。在近日舉行的一次媒體發(fā)布會(huì)中,DeVos表示,該項(xiàng)目的起始階段預(yù)計(jì)要持續(xù)至2019年,屆時(shí)將在新加坡的自動(dòng)駕駛車輛開發(fā)測(cè)試場(chǎng)地緯壹科技城投入6輛將沿固定路線低速行駛的改裝版量產(chǎn)汽車。
在試點(diǎn)項(xiàng)目的初始階段,工程師將作為“安全駕駛員”,陪伴乘坐這些自動(dòng)駕駛汽車的乘客。DeVos表示,該項(xiàng)目的第二階段“將采用一款真正可以實(shí)現(xiàn)‘按需服務(wù)’的專用自動(dòng)駕駛車輛”。本質(zhì)上說,第二階段的自動(dòng)駕駛車輛可隨時(shí)響應(yīng)客戶的召喚。DeVos說,當(dāng)項(xiàng)目完成后,我們就可以“展示一套完整而耐用的‘生態(tài)系統(tǒng)’,其中包括數(shù)據(jù)分析和最終用戶的反應(yīng)”。
今年晚些時(shí)候,德爾福將在北美地區(qū)新增一個(gè)試點(diǎn)項(xiàng)目,未來歐洲試點(diǎn)項(xiàng)目也將開始進(jìn)行。DeVos解釋道,第二階段的定制車輛支持車門自動(dòng)開合操作,可以方便殘疾乘客的上下車。
德爾福公司總裁兼CEO Kevin Clark在一份聲明中表示,AMoD項(xiàng)目將展示公司在自動(dòng)駕駛軟件、多模傳感器技術(shù)及系統(tǒng)整合方面的專業(yè)能力,并同時(shí)展現(xiàn)新加坡在互聯(lián)汽車研發(fā)和自動(dòng)駕駛基礎(chǔ)設(shè)施建設(shè)方面的領(lǐng)導(dǎo)地位。自2014年以來,新加坡自動(dòng)駕駛汽車計(jì)劃(SAVI)已經(jīng)與來自多個(gè)行業(yè)的合作伙伴建立合作關(guān)系,共同開展更多有關(guān)自動(dòng)車和試驗(yàn)臺(tái)方面的研發(fā)。
公司CTO Jeff Owens指出,新加坡的項(xiàng)目將借助德爾福“2015年自動(dòng)駕駛橫穿美國(guó)之旅”中使用的多項(xiàng)技術(shù)。具體來說,德爾福的這一項(xiàng)目借助一款改裝奧迪SQ5實(shí)現(xiàn)了橫穿美國(guó)全境的自動(dòng)駕駛之旅,是實(shí)現(xiàn)Owens所稱新型“終端到終端移動(dòng)解決方案”進(jìn)程中的一次巨大進(jìn)步。DeVos指出,那場(chǎng)橫穿美國(guó)之旅“吸引了新加坡的注意。”
新加坡道路交通管理局正在研究如何協(xié)助乘客更加方便地在住宅、公共交通站點(diǎn)和工作場(chǎng)所之間進(jìn)行日常通勤。他們認(rèn)為,通過為顧客提供“24/7的按需服務(wù)”自動(dòng)駕駛車輛,解決他們“第一英里”和“最后一英里”的通勤需求,可以提高公共交通系統(tǒng)的使用率,從而緩解新加坡的整體交通擁堵和汽車尾氣排放問題。
DeVos斷言,“由于前往公共交通站點(diǎn)不方便,所以人們出行時(shí)常常選擇出租車,而這會(huì)使新加坡的交通更加擁堵。”他補(bǔ)充說,按照德爾福的計(jì)劃,公司自動(dòng)駕駛車輛的服務(wù)對(duì)象不僅包括乘客,還可以運(yùn)送貨物。
德爾福在新加坡項(xiàng)目中規(guī)劃了三條受到嚴(yán)格控制的路線。目前,德爾福仍在尋找“五到六個(gè)”供應(yīng)商作為合作伙伴,其中包括負(fù)責(zé)云計(jì)算平臺(tái)的公司。DeVos表示,團(tuán)隊(duì)中可能包括軟件服務(wù)公司Mobileye。他說,這種努力“可能最終帶來一種以用途為中心的地圖服務(wù)。”
據(jù)了解,為長(zhǎng)約7到8公里的行車路線繪制精度在30厘米的地圖,預(yù)計(jì)需要4到6周的時(shí)間,其中包括數(shù)據(jù)匯編工作。
截止本文發(fā)表之時(shí),德爾福尚未透露首批測(cè)試汽車的具體廠商和車型。
“隨著AMoD項(xiàng)目的推進(jìn),未來的出行成本將顯著下降。”DeVos表示,“我們預(yù)計(jì),該項(xiàng)目可以證明我們的[自動(dòng)駕駛]技術(shù)非常穩(wěn)健,完全可以滿足消費(fèi)者的期望。”
作者:Lindsay Brooke
來源:SAE汽車工程雜志
翻譯:SAE中國(guó)辦公室

Delphi launches driverless pilot program in Singapore, aims for Level 4 operation by 2019
Delphi Automotive on August 1 announced an extensive autonomous-vehicle pilot program in Singapore, aimed at demonstrating cloud-based fully automated mobility on demand (AMoD) capability “at the [SAE] Level 4 performance level” by late 2019, said Glen DeVos, Vice President of Delphi’s Business Services Unit based in Mountain View, CA.
Operational capability is expected by 2022. Delphi is partnering with the Singapore government’s Land Transport Authority (LTA) on the multi-phase project. The initial phase, to be conducted through 2019, will involve a fleet of six modified production vehicles operating at low speed on fixed routes in the island nation’s "one north" area, a business park that is currently serving as a test bed for autonomous-vehicle development, DeVos explained in a recent media briefing.
Engineers acting as “safety drivers” will accompany select commuters during the first pilot phase. The program’s second phase “will use a true purpose-built, autonomous mobility-on-demand vehicle,” DeVos said—essentially driverless taxi-pods that can be summoned by customers. When completed the program “will show we have the complete ‘ecosystem’ and durability, including data analytics and reaction by the end consumer,” he said.
Delphi will announce an additional pilot in North America later this year and will also replicate the program in Europe. The second-phase bespoke vehicle will require automated door operation to easily accomodate passengers with physical disabilities, DeVos explained.
Delphi President and CEO Kevin Clark said in a statement that the AMoD project will demonstrate his company’s prowess in automated software, multi-modal sensor technology and systems integration while showcasing Singapore’s leadership in connected-vehicle and autonomous infrastructure. Since 2014 the Singapore Autonomous Vehicle Initiative (SAVI) has increased autonomous-vehicle research and test-bedding with various industry partners.
CTO Jeff Owens noted that the Singapore program will leverage technologies used in the first-ever coast-to-coast U.S. autonomous drive conducted by Delphi in 2015. That project used an Audi SQ5 platform and was a significant step in creating what Owens called “an end-to-end solution” for new mobility markets. The trans-U.S. drive “caught Singapore’s attention,” noted DeVos.
The Singapore LTA is studying ways to assist commuters in their daily round trip from home to mass-transit station to workplace. Offering on-demand automated vehicles on a 24/7 basis for what planners call “the first mile” and “last mile” of the typical commute would, they believe, increase use of mass transit systems and reduce overall traffic congestion and vehicle emissions in the process.
“It’s not easy to get to mass transit in Singapore so people take taxies, increasing congestion in the process,” DeVos asserted. He added that Delphi intends the service to include good and services in addition to people.
Delphi is conducting its own mapping for the three highly controlled routes included in the Singapore project. The company is still finalizing its “five or six” supplier-partners, including that for the cloud platform. The team could include Mobileye, said DeVos. The effort “could possibly lead to a production-intent mapping service,” he said.
Mapping the 7 to 8 km (4.3 to 4.9 mi) of each of the three routes to a 30-cm (12-in) level of accuracy consumed 4 to 6 weeks of time, including data compilation.
Delphi had not yet revealed OEM and models of the initial test-fleet vehicles at the time this article was published.
“With AMoD, the cost of the trip goes down significantly,” DeVos said. “We expect this project will prove our [autonomous] technology is robust and that consumers will use it.”
Author: Lindsay Brooke
Source: SAE Automotive Engineering Magazine