自2008年起，經濟衰退對整個行業起到了喚醒作用，汽車市場經歷了明顯的結構性變化。最初為智能手機開發的成像技術，現在已經侵入汽車電子，並逐步走向“中心舞台”。2015年，平均每輛汽車的攝像頭數量不足一個，而到2021年，預計平均每輛汽車將擁有3個以上攝像頭，這意味著汽車圖像傳感器的出貨量將達到3.71顆。
攝像頭最初安裝在高端汽車上，主要用於高級駕駛輔助系統（ADAS），具有深度學習圖像分析技術，以促進早期應用。以色列Mobileye公司致力於將該項技術推向汽車市場，硬件方面，其採用了安森美（ON Semiconductor）的CMOS圖像傳感器。隨著市場對資本的吸引，模仿競爭者可能會出現。現在可以確定的事實：基於視覺的自動緊急剎車系統（AEB）是可以實現的，並且能夠挽救生命。因此，前置ADAS攝像頭將加速發展和應用。
汽車成像市場的增長也受到停車輔助應用的驅動，360度全景攝像頭的出貨量暴漲。雖然美國要求強制安裝後視攝像頭，但這與360度全景攝像頭的“鳥瞰視角”相比，還是顯得微不足道。該趨勢有益於傳感器廠商，如索尼（Sony）和豪威科技（Omnivision），以及攝像頭製造商，如松下（Panasonic）和法雷奧（Valeo）。
攝像頭取代汽車傳統後視鏡仍不確定，該市場起飛主要取決於其外觀和汽車設計規則。歐洲和日本處於這一趨勢的前沿位置，但是2021年其對攝像頭整體市場的貢獻仍非常小。
固態激光雷達很受歡迎，預計2021年將在高端汽車中獲得應用。半自動/全自動駕駛推動激光雷達市場發展，但是降低成本是一個關鍵因素。本報告將詳細分析激光雷達對汽車視覺的影響。
採用長波紅外（LWIR）技術的夜視攝像頭最初是高端汽車的象徵。然而，現在它們越來越多地應用於行人和野生動物的自動檢測。因此，未來LWIR攝像頭將被整合到汽車ADAS系統中。
3D攝像頭主要用於車內信息娛樂和駕駛員監控。這項技術是豪華車的重要配置，因此如今的汽車搭載率仍非常有限。如果半自動駕駛的趨勢不斷升溫，那麼考慮到安全問題，該技術可能將成為強制性要求。
汽車ADAS傳感器
成像技術將汽車產業道路轉向自動駕駛模式
知名的矽谷公司，如穀歌、Uber和特斯拉，正在投資自動駕駛技術，讓消費者充滿期待和想像力。攝像頭和視頻反饋是一個明顯的技術標誌，正成為汽車製造商們區分自己的必備元素。應用於圖像顯示的攝像頭，如停車輔助，正以每年30%的速度增長。然而，這類攝像頭也將快速商品化，平均銷售價格（ASP）將快速降低，不過可以通過增加分辨率和攝像頭數量來緩解降價壓力。
現在，前置ADAS攝像頭市場的年增長率約為22%，三焦距攝像頭（tri-focal cameras）和立體安裝（stereo-mounting）正越來越受歡迎。新技術將提高安全性，推動早期的半自動功能應用。諸如攝像頭模組廠商、光學透鏡廠商之類的行業供應商充分挖掘上述技術驅動的生態環境。亞洲領導廠商，如舜宇光學和Sekonix，正在受益於該趨勢。
攝像頭不是與雷達或激光雷達競爭，它們三個都可以集成於一輛汽車，主要考慮因素是價格和自動駕駛級別。隨著汽車價格達到一定水平，攝像頭是最先被安裝於汽車的成像傳感器，主要原因是價格低廉。然後將是雷達，最後是激光雷達。
由於Delphi、Quanergy和SensL等公司的努力，激光雷達目前正在進行技術和產業變革。通過將其變成“固態”形式，促進價格下降，以期望進入汽車ADAS應用。現在，激光雷達逐漸被高端汽車市場接受，將成為半自動駕駛汽車的配備。
2012~2020年汽車攝像頭營收預測
紛繁複雜的技術路線將帶來許多機遇
早期的障礙已被克服，如寬動態範圍和大視野。新的問題正在認真解決中，如LED閃爍和低光照問題。索尼（Sony）和意法半導體（STMicroelectronics）最近發布的汽車傳感器是推動技術進步的案例，也希望取代安森美（ON Semiconductor）在汽車ADAS成像領域的領導地位。CMOS圖像傳感器技術是當前成熟度最高且廣泛應用的成像技術，技術的進步速度也非常快，諸如移動設備中重要的創新技術也可能發生在汽車圖像傳感器中。
應用的多樣性和傳感器數量的增加使得產品專業化，各種市場機遇湧現，包括常規的攝像頭、3D和紅外攝像頭、激光雷達等。反過來，圖像處理技術也需要充分利用傳感器的成像能力。多種處理方法正在進入市場，如NVidia的圖形處理器（GPU）、Kalray的many-core等。隨著汽車電子的蓬勃發展，成像技術已經成為不可或缺的汽車“感覺器官”。
汽車成像技術路線圖
Imaging technology, which is currently mainly cameras, is exploding into the automotive space, and is set to grow at 20% CAGR to reach $7.3B in 2021
INFOTAINMENT AND ADVANCED DRIVER ASSISTANCE SYSTEMS (ADAS) PROPEL AUTOMOTIVE IMAGING
Since 2008, when a recession acted as a wakeup call to the whole industry, the automotive market has undergone obvious structural change. Capitalizing on technologies initially developed for smartphones, electronics have invaded, and imaging technology is now taking center stage. From less than one camera per car on average in 2015, there will be more than three cameras per car by 2021, which means 371 million automotive imaging devices.
Cameras were initially mounted for ADAS purposes on high-end vehicles, with deep learning image analysis techniques promoting early adoption. The Israeli company Mobileye has been instrumental in bringing this technology to market, along with On Semiconductor, which provided the CMOS image sensor. Copycat competition will probably pick up as the market now justifies initial investment in design and technology. It is now a well-established fact that vision-based autonomous emergency braking (AEB) is possible and saves life. Adoption of forward ADAS cameras will therefore accelerate.
Growth of imaging for automotive is also being fueled by the park assist application, and 360° surround view camera volume is skyrocketing. While it’s becoming mandatory in the US to have a rearview camera, that uptake is dwarfed by 360° surround view cameras, which enable a “bird’s eye view” perspective. This trend is most beneficial to companies like Omnivision at sensor level and Panasonic and Valeo, which have become the main manufacturers of automotive cameras.
Mirror replacement cameras are currently the big unknown and take-off will primarily depend on its appeal and car design regulation. Europe and Japan are at the forefront of this trend, which should become slightly significant by 2021.
Solid state lidar is well talked about and will start to be found in high end cars by 2021. Cost reduction will be a key driver as the push for semi-autonomous driving will be felt more strongly by car manufacturers. The report will analyse the impact of lidar for automotive vision in detail.
Night vision cameras using Long Wave Infrared (LWIR) technology were initially perceived as a status symbol. However, they’re increasingly appreciated for their ability to automatically detect pedestrians and wildlife. LWIR will therefore become integrated into ADAS systems in future.
3D cameras will be limited to in-cabin infotainment and driver monitoring. This technology will be key for luxury cars and therefore is of limited use today.
If any significant semi-autonomous trend picks up, the technology will probably become mandatory, due to safety issues.
IMAGING WILL TRANSFORM THE CAR INDUSTRY EN-ROUTE TO THE SELF-DRIVING PARADIGM SHIFT
Highly renowned Silicon Valley companies such as Google, Uber, and Tesla are investing in autonomous driving, bolstering consumer imagination. Cameras and video feedback is an obvious technology marker, which is becoming a must-have for car makers to differentiate themselves. Cameras for display purposes such as park assist are growing at 30% annually. However, they will suffer from quick commoditization, although lower average selling prices (ASPs) will be mitigated by increased resolution and increased volume.
Forward looking ADAS cameras are growing 22% annually will get more complex, which will keep prices high. Today tri-focal cameras and stereo-mounting are gaining popularity. New technologies will raise the bar in terms of security and drive the adoption of early semi-autonomous features. Industrial providers such as camera module players and optical lens set players are capitalizing on this technology-driven environment. Asian leaders such as Sunny Optical and Sekonix are clearly benefiting from this trend.
Cameras don’t really compete with radar or lidar, and all three could be integrated, depending on price and the level of autonomy targeted. As car price reaches certain levels, cameras are the first to be mounted due to their low ASP, then radars are added, then eventually lidars.
Lidars are currently undergoing a technological and industrial revolution, thanks to companies like Delphy, Quanergy and SensL. By becoming solid-state they should be considered as true imaging devices. Their price point will fall below radars’, which are commonly used for ADAS, and consequently they will progressively be adopted by the top end of the automotive market. They will become central to semi-autonomous featured vehicles.
A MAZY TECHNOLOGICAL ROADMAP WILL BRING MANY OPPORTUNITIES
Early hurdles such as wide dynamic range and large fields of view have been overcome. New issues such as LED flickering and low light issues are currently being considered seriously. SONY and ST Microelectronics’ recent automotive sensor release are examples of the push for technology excellence aimed at displacing ON
Semiconductor leadership in ADAS image sensors.
The current maturity of CIS technology allows for wide adoption of imaging technologies and is also accelerating research for technical efficiencies. Improvements will drive competition and major breakthroughs like the innovations seen in mobile devices could also happen in vehicles image sensors.
The diversity of applications and the sheer numbers of devices, including conventional, 3D and IR cameras and lidars, allows specialization. Image processing can in turn exploit the devices’ capabilities. Multiple processing approaches such as NVidia GPUs, or Kalray’s many-core approaches are competing to enter the market. As automobiles undergo electronic metamorphoses, imaging technology has become a key enabling technology.
- Source:YOLE
留言列表
生技醫療產業 (1)