Artificial intelligence that blows the wind, autonomous driving that makes money lying down | Tiger

These days, if you don’t talk about artificial intelligence when starting a business, you won’t get any money at all.

Large companies engage in artificial intelligence, executives of large companies leave to start their own businesses to engage in artificial intelligence, and investors invest in artificial intelligence. Artificial intelligence companies have undoubtedly become a money-attracting pool among Internet companies. Not to mention reliable companies, as long as they have reliable employees, they can get a large amount of entrepreneurial capital, which is enough to ride the tide of artificial intelligence for a period of time.

But to be more practical, the current research directions of artificial intelligence in these technology companies are actually three main ones: deep learning, natural speech processing, and machine vision. Deep learning is a required course for artificial intelligence, natural language processing can directly apply technology to virtual intelligent assistants, and machine vision can be directly transformed into autonomous driving technology.

Smart speakers equipped with virtual voice assistants have been around for a long time, but judging from consumer feedback, they do not think smart voice assistants are smart things, and even smart speakers currently look a bit useless. Unlike smart speakers, which are hot in the market but are cold among consumers, autonomous driving seems to be a project that both investors and ordinary consumers are optimistic about. Many people expect autonomous driving to improve future traffic conditions and solve the current congestion and inefficient traffic problems. The capital market also favors autonomous driving companies and autonomous driving talents.

This morning, self-driving startup Momenta announced a $46 million Series B investment. This round of financing was led by NIO Capital, with participation from Daimler Group (parent company of Mercedes-Benz), Shunwei Capital, Sinovation Ventures and Jiuhe Venture Capital. In 2016, Momenta received Series A financing led by Blue Lake Capital, with participation from Sinovation Ventures and Zhen Fund. In early 2017, it received Series A1 financing led by Shunwei Capital.

The company has raised multiple rounds of funding in less than a year since its establishment. The B round of 46 million is indeed a considerable investment, especially since the company currently has no external output.

Let’s dig deeper into this company.

Momenta is a Beijing-based autonomous driving company established in September 2016. Founder Cao Xudong previously worked at Microsoft Research Asia and SenseTime. The company's goal is to develop the brain of autonomous driving. Its core technologies are environmental perception, high-precision maps, and driving decision-making algorithms based on deep learning. Products include different levels of autonomous driving solutions and derived big data services.

In other words, Momenta is an autonomous driving solutions company.

Compared with other companies, Momenta’s technical advantage lies in visual recognition. According to Cao Xudong, Momenta has the world's top deep learning experts, the authors of the most advanced frameworks in the field of image recognition, Faster R-CNN and ResNet, and multiple competition champions in ImageNet 2015 and MS COCO Challenge 2015. The team comes from Tsinghua University, MIT, Microsoft Research Asia, etc., and has profound technical accumulation and strong technical originality. The company’s R&D director, Ren Shaoqing, is the inventor of Faster RCNN, the world’s most widely used object detection framework.

Such advantages in visual recognition technology have given the company a relatively large advantage in the most basic perception part of the autonomous driving solution. Because from the current point of view, visual recognition will become the main solution for the perception part of autonomous driving. Compared with lidar, which is widely used in the early stages of autonomous vehicles, the cost of camera sensors is lower, and the data generated is smaller than the data generated by lidar mapping of the surrounding environment. It is currently a more practical solution for realizing autonomous driving or driving assistance.

However, the difficulty of visual recognition lies in the high-precision visual recognition algorithm. Only if the algorithm is optimized well, the on-board computer can identify and process the images collected by the camera and divide the drivable areas, pedestrians, street lights and other road traffic elements.

And algorithms are what Momenta is good at.

However, to achieve autonomous driving, good perception alone is not enough. Decision-making and control are the most critical technical modules for autonomous driving. In order to adjust the decision-making algorithm of autonomous driving, a large amount of actual road test data is needed to optimize the algorithm to achieve ultimately more intelligent and safer autonomous driving control.

However, accumulating autonomous driving data is a costly task, and it is difficult for small companies to do it, so it requires huge financial and human support.

So you can go back and see where the $46 million in financing will be used. Cao Xudong said: "The financing will be used in three aspects: 1. Strengthening the core capabilities of artificial intelligence, including big data, big computing and outstanding AI talents; 2. Productizing vision-based environmental perception and high-precision map technology; 3. Research and development of L4 driverless technology for high-frequency rigid demand scenarios."

At the current stage when various companies are accumulating autonomous driving technology, talent is the most expensive investment, because each company is wooing autonomous driving technical talents, causing the income of talents in this area to rise. Many large companies are even willing to invest hundreds of millions of dollars in order to acquire a small autonomous driving team established by a few technicians.

On the other hand, for self-driving startups, capital will not allow long-term investment in unproductive consumption. Everyone thinks that self-driving will make money in the future, but how much money can be made from self-driving and how to earn this money when self-driving is not yet popular are still difficult problems. Therefore, startups still need to gradually commercialize their technology and give investors enough confidence and company prospects before they can continue to invest in developing their own technology. For an autonomous driving company like Momenta that specializes in machine vision, the initial productization plan is likely to provide technical support for driving assistance to car manufacturers to make profits (investors can also see that Mercedes has also participated in the investment, also to make up for its technical weakness in visual recognition of driving assistance. You must know that Mercedes-Benz's research and development of driving assistance is not later than Tesla, but the visual recognition is far behind Tesla, which is why Mercedes-Benz's high-level driving assistance seems to have been tepid in the consumer market).

Of course, the ultimate goal of each company is to achieve L4 level fully autonomous driving. The target limit given by Momenta is L4 driverless technology for high-frequency rigid demand scenarios. In response to my country's traffic conditions, it is likely to be low-speed autonomous driving during traffic jams.

Nowadays, most startups are aiming for L4-level fully autonomous driving. Such ambitious goals are also particularly attractive to investors. However, autonomous driving is not something that can be solved in a short time. The 2020 or 2021 previously expected by major car manufacturers is definitely not a year when autonomous driving can be completely profitable. Investors love to play with the trend. How long can the wind of artificial intelligence and autonomous driving continue to blow? It depends on how patient the investment circle is.