越來越多的業(yè)內人士意識到開放平臺式軟件是快速、低成本部署自動駕駛汽車的捷徑,這也讓Renovo針對按需自動出行(AMoD)所研發(fā)的AWare操作系統(tǒng)獲得了越來越多的合作伙伴。
如果要列一張實現(xiàn)高級自動駕駛所需的技術清單,上面都會有哪些技術呢?首先自然是傳感器、計算機棧、車輛本身等硬件,當然也包括繪圖、人工智能、傳感器整合等復雜的軟件系統(tǒng)??偠灾@會是一張不短的清單。
面對如此繁多的自動駕駛技術,我們又該如何進行統(tǒng)籌整合呢?很多人都想當然地認為,答案就是建立一個類似于電腦或手機的操作系統(tǒng)。
然而在位于加利福尼亞州坎貝爾市的Renovo Auto公司看來,答案沒有這么簡單。Renovo Auto的CEO Chris Heiser認為,雖然目前有無數(shù)的開發(fā)機構在竭力完善各自的自動駕駛解決方案,其中也涵蓋了操作系統(tǒng),但是最終,自動駕駛汽車會像計算機行業(yè)一樣,只留下幾個標準操作系統(tǒng)平臺。
Heiser表示,最后留下來的一定有Renovo的開放平臺解決方案AWare。至少在針對按需自動出行(AMoD)SAE四級自動駕駛車隊的操作系統(tǒng)中,Renovo一定會勝出。
Heiser在接受《汽車工程雜志》的采訪時表示,“我并不是,說其它的操作系統(tǒng)的開發(fā)企業(yè)現(xiàn)在就會直接宣布放棄,然后說‘好吧,咱們還是買Renovo的技術吧。’”
“但是最終,他們一定會這么做,”Heiser充滿自信地補充道,“因為這是最明智的做法。” Renovo堅信,最高效的代碼是一次寫完后就可以融入不同的代碼庫和車型。像Waymo和通用Cruise 這樣的大型開發(fā)機構使用的是獨家操作系統(tǒng)架構,因此無法實現(xiàn)Renovo的效率和可擴展性。
Heiser表示,Renovo相信AMoD服務將徹底改變城市的客運和貨運。“剛開始AMoD的部署會比較受限,但是它會迅速擴展,這也是為什么我們如此看好這一細分市場。”
和志趣相投的合作伙伴一起成長
Renovo的合作伙伴都是大名鼎鼎的硅谷技術公司和底特律車企。激光雷達巨頭Velodyne、三星、自動駕駛技術集成商巨頭Aptive、網絡安全專家Argus都已在2017年成為了Renovode 合作伙伴。今年,Renovo又和人工智能開發(fā)企業(yè)Perceptive Automata公司及新起之秀Voyage建立了合作關系。Voyage正在建立一支搭載AWare系統(tǒng)的克萊斯勒Pacifiica自動駕駛車隊,屆時將在加利福尼亞州和佛羅里達州推出基于地理圍欄的AMoD服務。
Perceptive Automata的CEO兼聯(lián)合創(chuàng)始人Sid Mistra在一則公告中表示,“我們之所以選擇AWare操作系統(tǒng)平臺,是因為Renovo一直在不懈擴展部署,現(xiàn)在有越來越多的車隊開始采用AWare。我們和Renovo秉持著相同的愿景,都希望能建立一個可以無縫融入最先進的技術的生態(tài)系統(tǒng),在現(xiàn)實世界里大規(guī)模部署安全的自動駕駛解決方案。”
Heiser深知,Renovo和合作伙伴都非常需要彼此。和其它的用戶端產品一樣,AWare的成功取決于產量。
Heiser表示,“就像其它基于使用計費的授權軟件一樣,只有量產才能獲得可觀的收入。我們現(xiàn)在要做的就是和優(yōu)秀的技術使用者一起部署,不懈地搶占看好的市場,和我們的客戶一起成長。”
“坦白說,我想大家都是一樣的。每一家激光雷達企業(yè)、每一家硅谷公司、每一家自動駕駛AI公司只有實現(xiàn)量產才能繼續(xù)壯大。如果單靠賣出一件十億美元的產品,公司是無法盈利的。要想賺錢,必須以合理的價格實現(xiàn)規(guī)模銷售。”
開放平臺 = 最明智的選擇
Heiser表示,“我們認為操作系統(tǒng)的開發(fā)就是一個選擇縱向開發(fā)還是橫向開發(fā)的問題。幾乎所有你能想到的開發(fā)企業(yè)都在進行垂直整合,想要全面發(fā)展一項技術。但是我們想要開發(fā)的是一個可以橫向擴展的簡易平臺,其中的原理和一個很神奇的事物很像— —那就是互聯(lián)網。戴爾發(fā)明了處理器,思科搭建了網絡,甲骨文開發(fā)出數(shù)據庫,互聯(lián)網因此得以大規(guī)模發(fā)展。我們不應該選擇縱向開發(fā),因為如果要建立大型系統(tǒng),縱向開發(fā)的效率很低,而且風險很高。”
自動駕駛系統(tǒng)的開發(fā)和計算機系統(tǒng)難免存在相似點,畢竟自動駕駛操作系統(tǒng)開發(fā)的本質也是尋求模塊化。Heiser認為,自動駕駛操作系統(tǒng)和計算機軟件一樣,是需要通過實際應用來進行證明的。
Heiser表示,“在硅谷呆了這么長時間,我們知道一個操作系統(tǒng)是如何變成標準操作系統(tǒng)的。我們目睹了微軟是如何成為臺式電腦的標準操作系統(tǒng)、Linux又是如何成為服務器運算的標準操作系統(tǒng);我們也見證了安卓是如何成為手機標準操作系統(tǒng)、AWS又是如何成為建立網絡服務的標準操作系統(tǒng)。每一套標準操作系統(tǒng)背后都有自己的故事,但是它們也有共同點,它們都是成本最低、操作最簡便、最易于擴展的系統(tǒng)。”
Heiser最后表示,眾所周知,計算機操作系統(tǒng)的成本已經大幅下降,而 AWare操作系統(tǒng)希望能讓自動駕駛操作系統(tǒng)的研發(fā)重演這段歷史。
他補充道,事實上,計算機和手機行業(yè)已經證明“橫向”開發(fā)往往優(yōu)于“縱向”開發(fā)。
Heiser聲稱,“Linux之所以能戰(zhàn)勝STUN、FGI和IBM、Digital、克雷等幾十個想要進行縱向開發(fā)的公司,是因為Linux是開放的、可擴展的,開發(fā)者可以很方便在Linux系統(tǒng)上推進工作。安卓也是以同樣的原因取代了當時諾基亞、三星、HTC、LG和三洋電機,結束了它們的重復性開發(fā)工作。”
現(xiàn)在輪到AWare成為自動駕駛開發(fā)者的標準操作系統(tǒng)了。Heister表示,雖然AWare目前針對的是AMoD車隊,但是AWare最終也會適用于自動駕駛私家車。不過他也承認,離這一天的到來還需時日,自動駕駛私家車的“運行包線”涉及到更加復雜的路徑問題。
但是在市區(qū)或其它明確的地理圍欄區(qū)域,Renovo已經準備好在面向公眾和商用車車主的自動駕駛出行服務中一展拳腳。
Heiser興奮地說道,“如果大家都準備好放棄開車,使用AMoD服務,那么我們離AWare成為標準操作系統(tǒng)的那一天就不遠了。”
Renovo’s AWare operating system for AutomatedMobility on Demand (AMoD) is expanding its reach as more players see open-platform software as a unifying—and simplifying—answer to quicker and less-costly AV deployment.
Consider a list of all the things high-level driving automation requires—hardware such as sensors, a computing stack, the vehicle itself. Software for a multitude of complex functions like mapping, artificial intelligence, sensor integration. The list would be exhaustive.
What will lord over it, see that everything integrates, make sense of it all? Much like a laptop computer or cell phone, an operating system, of course.
But that simple conclusion isn’t—at least from the standpoint of Campbell, California’s Renovo Auto. With countless developers toiling to perfect their specific ingredients for the automated-driving recipe, including operating systems, Renovo CEO Chris Heiser believes that just like the computer industry, automated-driving development will converge to the use of just a couple of standard operating-system platforms.
He further predicts Renovo’s open-platform solution, AWare, is destined to win. At least for the automated-mobility-on-demand (AMoD) fleet-use application of SAE Level 4 automation Renovo is targeting.
“I’m not saying that every single one of the groups building internal operating systems is just going to kind of declare defeat and say, ‘Hey, we want to license Renovo’s stuff,’” Heiser said in an interview with Autonomous Vehicle Engineering.
“But eventually they will,” he added with no small degree of confidence, “because that’s the best way to do this.” Renovo is convinced of the efficiency of writing code once and integrating across a widespread code base and vehicle parc. Mega-developers such as Waymo and GM Cruise, using their own bespoke OS architectures, won’t achieve the same efficiency and scale capability, Heiser insisted.
Heiser said Renovo is certain AMoD services are ready to drastically change transportation of people and cargo in urban areas. “The deployments will be initially limited, but they will grow quite quickly and that’s why we’re so bullish about this segment,” he said.
Growing with like-minded partners
Renovo’s list of collaborators already is a who’s-who of Silicon Valley and Detroit names: lidar colossus Velodyne, Samsung, auto-technology mega-integrator Aptiv, cybersecurity expert Argus all threw in with Renovo in 2017. This year saw partnerships inked with artificial-intelligence developer Perceptive Automata and rising-star Voyage, which is working now to integrate AWare for its fleet of automated Chrysler Pacificas that will operate in geo-fenced AMoD service in California and Florida.
“We chose to integrate with the AWare OS platform because of Renovo’s focus on deployment at scale and access to the growing number of fleets running AWare,” Sid Misra, CEO and Co-Founder of Perceptive Automata, said in a release. “We share Renovo’s vision of an ecosystem where best-in-class technologies seamlessly integrate to enable the safe large-scale deployment of automated mobility solutions that are suitable for the real world.”
Heiser knows Renovo needs its partners as much as they need Renovo. As with almost all forms of consumer-reaching products, success is tied to volume.
“Just like any other usage-based software licensing, volume is the thing that
drives massive revenue,” he said. “The name of the game for now, for us, is make sure that we’re deploying with great customers, make sure that we’re aggressively going after markets that we believe in and grow as our customers grow.
“And I think frankly that’s true to everyone in the space. Every lidar company, every Silicon company, every self-driving AI company, we all grow only with volume. No one makes any money by selling one of these things for a billion dollars; you make money selling lots of them for a reasonable amount of money.”
Open-platforms = widest adoption
‘The way we think of (OS development) is just simply vertical versus horizontal,” explained Heiser. “Almost every other group you can think of is vertically integrating and they’re responsible for pretty much everything. Instead, we are focusing on a thin layer that is horizontally scalable—and by the way, there’s this awesome thing called the Internet which happened exactly like this. Dell, with processors. Cisco did networks. Oracle did databases. That’s how you get to scale. You don’t go vertical, it’s a very inefficient and risky way to build large systems.”
There are inevitable computer-industry analogies because OS development is, at its foundation, much like that of any other business that seeks modularization. Moving into the automated-vehicle space is merely the latest application for tried-and-true computer-software maxims, Heiser explained.
“Being in the Silicon Valley for a long time, we’ve seen how this happens. We watched how Microsoft became the standard operating system for desktop, how Linux became that for server computing, how Android became that for mobile, how AWS became that for building web services,” Heiser said. “And each one of their stories is different, but the common thread is that they become the least expensive and easiest way to build things at scale.”
The AWare operating system is, Heiser concludes, nothing more than an example of bringing to autonomy development what played out in the computer industry— which is famous for its history of cost-cutting magnitude.
In effect, he added, the computer and mobile-phone sectors proved what is “vertical” is typically better when it becomes “horizontal.”
“Linux replaced what IBM and STUN and Digital and FGI and Cray and a dozen in tandem and a dozen other companies were all doing vertically by themselves,” Heiser said. “And it replaced it because it was open, it was scalable and it was a great way for developers to push out what they were working on. Same thing with Android: Android replaced what Nokia and Samsung and HTC and LG and Sanyo were all doing duplicatively in-house.”
Now it’s time for that same revolution to come to autonomy developers—for now, at least, for AMoD fleets. But eventually, he said, AWare is absolutely applicable for governing automation systems for personal vehicles, too. Heiser admits, however, that day is longer coming, calling the “operational envelope” for personal autonomous vehicles a more difficult development path.
But for those in urban environments and other defined, geofenced areas, Renovo is prepped for an imminent cascade of automated-vehicle service for individuals and for commercial users.
“I’m ready. I’m ready to give up my car,” Heiser effused. “I’m ready to utilize this service. Then, the big capital is going to come.”
Author: Bill Visnic
Source: SAE Automotive Engineering Magazine