绝命毒师

Pony.ai’s autonomous-driving distance tops 100m km as China scales up L4 commercialization_我的网站

一 | 北京时间1月16日，据外媒报道，每个人都知道达芬奇名画《蒙娜丽莎》中的女子似乎在回望观众的眼睛，无论观众在哪里看这幅画，蒙娜丽莎的眼睛似乎都在跟着自己。但事实并非如此。    

    Chinese autonomous driving firm Pony.ai launched commercial robotaxi services in Zagreb, Croatia on April 8. Photo:Courtesy of Pony.ai
Chinese autonomous-driving company Pony.ai said on Monday that its cumulative autonomous-driving distance worldwide has surpassed 100 million kilometers, marking the latest milestone as China's driverless sector moves from technology validation toward mass production and scaled commercial deployment.
The milestone comes as L4 autonomous driving gains commercial traction across passenger transport and logistics in China, while Chinese companies are accelerating overseas deployment through partnerships with local operators in markets including Europe and Southeast Asia.
Industry experts said that the significance goes beyond rising distance. As commercialization gathers pace, the next phase of competition will increasingly hinge on whether autonomous-driving technology can be deployed at scale, brought down to commercially viable costs and replicated across different transport scenarios and markets.
At a recent media briefing, Pony.ai said that its fourth-generation L4 autonomous electric heavy-duty truck had entered mass production, with commercial operations expected to begin in the coming months. The company plans to deploy 500 to 1,000 trucks over the next two to three years across long-haul freight, bulk cargo transport and port logistics.
A notable part of that transition is the sharp decline in costs. Autonomous-driving hardware costs for Pony.ai's fourth-generation heavy truck are about 70 percent lower than for the previous generation, according to the company.
Similar cost reductions are also emerging in robotaxis. Pony.ai said on its official blog that the bill-of-materials cost of the autonomous-driving kit for its seventh-generation robotaxi has fallen 70 percent from the previous generation, helping the company achieve unit-economics breakeven in Guangzhou and Shenzhen, South China's Guangdong Province.
China's autonomous-driving sector is moving from technology validation to large-scale commercial deployment, underpinned by the rapid accumulation of real-world driving data, Pan Helin, a member of the Expert Committee for the Ministry of Industry and Information Technology, told the Global Times on Monday. 
"Continuous road testing allows L4 systems to better adapt to local traffic conditions, laying the groundwork for wider commercial use," Pan said.
China has opened multiple cities to autonomous-driving applications, with commercialization gathering pace. As deployment expands and marginal technology costs fall, autonomous driving is expected to unlock significant commercial value and emerge as a new form of mobility, veteran industry analyst Liu Dingding told the Global Times.
Pony.ai aims to expand its robotaxi fleet to more than 3,000 vehicles by the end of 2026 and operate in 20 cities worldwide, with nearly half of those cities overseas. It also plans to develop an international version of its seventh-generation robotaxi tailored to different regulatory requirements and infrastructure, supporting the deployment of more than 1,000 vehicles in overseas markets. 
Pony.ai's ambition also reflects a broader push by Chinese autonomous-driving companies to expand overseas.
On August 3, WeRide announced a partnership with Danish shared-mobility platform GreenMobility to deploy autonomous ride-hailing services in Denmark, marking its first entry into the Nordic region and sixth European market. Subject to regulatory approval, the service is expected to open to the public in the first half of 2027 using WeRide's latest GXR L4 robotaxi.
WeRide said on its website that its autonomous-driving products are now deployed in more than 40 cities across 12 countries. The company holds autonomous-driving permits in eight markets and operates a global L4 fleet of more than 3,000 vehicles.
A similar expansion is taking place in public transport. On July 17, Chinese autonomous-driving company MOGOX announced a strategic partnership with Singapore-based ComfortDelGro Bus. The two companies will cooperate on autonomous-vehicle deployment and operations, commercial expansion and coordination with regulators.
Pan said that China's autonomous-driving sector has two key advantages: a broad supplier base for major components and a large domestic market that allows companies to deploy services widely and accumulate real-world data. Advances in simulation and synthetic-data technologies have further strengthened the sector's ability to train and refine autonomous-driving systems.
Liu pointed to the cost benefits of scale. As deployment expands, the cost of key automotive-grade components such as lidar and chips continues to fall, improving the economics of commercial operations and creating more room for technological iteration, he said.
。

二 | 蒙娜丽莎的眼睛实际上并不跟随周围的观察者。一项新的研究发现，蒙娜丽莎的目光实际上落在观察者的右侧15.4度，远远超出了通常让你觉得有人在盯着我的角度。德国bilefield大学的知觉心理学家、该研究的作者Gernot Horstmann说：“她没有在看你。”不真实的“蒙娜丽莎效应”有点讽刺，因为肖像或照片中的人似乎跟随观察者的运动的现象被准确地称为“蒙娜丽莎效应”。霍斯曼说这种效应确实存在。如果肖像或照片中的人直视前方，即使有人站在某个角度看照片，他或她也会感到肖像或照片中的人正盯着自己。只要观察者偏离不超过5度的角度，蒙娜丽莎效应就会发生。这对观众与屏幕人物的互动具有重要意义。如果你想让坐在房间右侧的观众感觉屏幕上的角色在看自己，不要让角色朝那个方向看，否则观众会觉得角色没有在看房间里的任何人。为了达到这个效果，人们应该直视前方。Horstmann与合作作者Sebastian Loth一起研究了这种效应在创造人工智能“替代物”中的应用。在这个过程中，霍斯曼长时间地看着蒙娜丽莎，终于意识到了一个问题。”我意识到：“等一下，她没有盯着我看。为了确保这不是为了他们自己，研究人员让24名受试者在电脑屏幕上观看蒙娜丽莎的照片。他们在观看者和屏幕之间放置了一把尺子，让受试者指出尺子和蒙娜丽莎眼睛交叉处的数字。为了测试照片中的其他元素是否影响观众对蒙娜丽莎眼睛的感知，研究人员放大了镜头。有时观众可以看到蒙娜丽莎的整个脸，有时只能看到眼睛和鼻子。

三 | 另外，为了计算蒙娜丽莎眼睛的角度，研究人员在实验中途改变了尺子和屏幕之间的距离。这样就可以得到两个交点，从而计算出角度。《蒙娜丽莎》的研究结果表明，所有受试者都认为《蒙娜丽莎》中的人物并不是直接看自己，而是稍微看自己的右边。”“蒙娜丽莎的观点显然超出了你通常会觉得有人盯着你看的感觉。”霍斯曼指出。那么，为什么人们总是相信蒙娜丽莎的眼睛会随着她的观察者四处移动呢？霍斯曼对此并不确定，但他相信人们都有关注的欲望，所以他们认为照片中的女人正盯着自己。也许是因为最初提出“蒙娜丽莎效应”这个词的人认为这很酷。

Current article:http://www.roubutianquegumaigoubingdian.cyou/list_6e4/injp.html

Published on:10:16:58