以自動駕駛為驅(qū)動力，視覺算法在車載領(lǐng)域找到了最佳落點(diǎn)。

今年4月舉行的上海車展上共有210款車型、1413臺車輛參展，其中以智能座艙和智駕輔助為主的新能源汽車占據(jù)絕對C位，尤其是屏幕以及攝像頭數(shù)量激增，讓整車科技感十足。如問界M9滿配版全車配備10塊屏幕，理想汽車所展示的L7、L8、L9全車內(nèi)外搭載了10顆以上攝像頭。

不止是新能源電動系，傳統(tǒng)燃油車也在不斷引入智能化技術(shù)，通過搭載視覺感知設(shè)備與雷達(dá)配備了多種行車輔助系統(tǒng)，如自適應(yīng)巡航、盲點(diǎn)監(jiān)測、車道保持等功能，并同時引入了智能語音控制、視覺感知交互、遠(yuǎn)程診斷和控制、車載娛樂等功能，使得車輛更加便捷和智能化。

通過視覺算法的不斷發(fā)展和優(yōu)化，汽車自動駕駛系統(tǒng)正帶來更加安全與舒適的駕乘體驗(yàn)，比如精準(zhǔn)實(shí)現(xiàn)障礙物檢測和跟蹤、車道檢測和跟蹤、路面狀況檢測等功能。

在智能電動汽車和自動駕駛的新時代，曾經(jīng)的視覺算法霸主Mobileye逐漸掉隊(duì)，眾多同類國產(chǎn)企業(yè)如雨后春筍冒出，紛紛欲取而代之。

根據(jù)潮電智庫了解，目前終端車企更青睞于“軟硬結(jié)合”的供應(yīng)模式。這意味著除了傳統(tǒng)的軟件算法支持，整合上下游資源，整體系統(tǒng)集成是對車載視覺方案商新能力的考量。

華為、虹軟、佑駕創(chuàng)新、瞰瞰智能科技、元橡科技、歐菲光等企業(yè)經(jīng)過多年的積聚與沉淀，已經(jīng)具備較強(qiáng)的綜合競爭力。

值得注意的是，自從激光雷達(dá)登上車載舞臺，與視覺算法方案的比較就不曾停止。

潮電智庫認(rèn)為，激光雷達(dá)和視覺算法應(yīng)該是相輔相成的關(guān)系。事實(shí)上，包括造車新勢力在內(nèi)的中國汽車廠商幾乎都采用了多傳感器融合路線。

不管怎樣，視覺算法在未來自動駕駛領(lǐng)域依然是主流的核心技術(shù)之一，其低成本、高成熟度與應(yīng)用廣泛性等優(yōu)勢無法替代。

Top 20 domestic vision algorithm solution providers in 2023

Driven by autonomous driving, visual algorithms have found the best landing point in the automotive field.

At the Shanghai Auto Show held in April this year, a total of 210 models and 1413 vehicles were exhibited, with new energy vehicles mainly featuring intelligent cabins and intelligent driving assistance occupying the absolute C-spot. Especially with the rapid increase in the number of screens and cameras, the entire vehicle has a strong sense of technology. The full version of the Wenjie M9 is equipped with 10 screens, and the L7, L8, and L9 displayed by the Ideal Car are equipped with more than 10 cameras both inside and outside the car.

Not only new energy electric vehicles, but traditional fuel vehicles are also constantly introducing intelligent technology. They are equipped with various driving assistance systems such as adaptive cruise control, blind spot monitoring, lane keeping, and other functions through visual perception devices and radar. At the same time, intelligent voice control, visual perception interaction, remote diagnosis and control, and in car entertainment are also introduced, making the vehicle more convenient and intelligent.

Through the continuous development and optimization of visual algorithms, the auto drive system is bringing a safer and more comfortable driving experience, such as accurate obstacle detection and tracking, lane detection and tracking, road condition detection and other functions.

In the new era of intelligent electric vehicles and autonomous driving, the once dominant visual algorithm, Mobileye, has gradually fallen behind, and many similar domestic enterprises have sprung up like mushrooms, eager to replace it.

According to the Electrend, terminal car companies currently prefer a supply model that combines software and hardware. This means that in addition to traditional software algorithm support, integrating upstream and downstream resources, the overall system integration is a consideration of the new capabilities of car vision solution providers.

After years of accumulation and accumulation, companies such as Huawei, Arcsoft, Minieye, KANKAN, METOAK, and Ofilm have already possessed strong comprehensive competitiveness.

It is worth noting that since the launch of LiDAR on the vehicle stage, the comparison with visual algorithm solutions has not stopped.

Electrend believes that LiDAR and visual algorithms should complement each other. In fact, almost all Chinese car manufacturers, including new car manufacturers, have adopted a multi-sensor fusion route.

Anyway, visual algorithms will still be one of the mainstream core technologies in the future field of autonomous driving, and their advantages such as low cost, high maturity, and widespread application cannot be replaced.