Total seventy-fourth (April 2018 visit to Beijing) Author Yi Xiao Yi After two years of entrepreneurship, Liu Shaoshan has kept his diary, weekly diary and his habit of thinking about writing. He has worked at the Baidu American Institute, the LinkedIn, Intel, and Microsoft Research Institute, focusing on the architecture and development of the unmanned vehicle system architecture and product, deep learning, and heterogeneous computing platforms. With the advantages of SLAM and other AI visual technology fields, Perceptln has created a single, binocular and four mesh visual computing module for different scenes, and provides an integrated software and hardware integration scheme for the robot. At present, these modules and solutions are applied in the field of sweeping robots, parks, driverless vehicles and other fields, and the cost is lower than that in the market. Figure: Perceptln binocular vision inertial computing module According to Liu Shaoshan, the focus of this years Perceptln is to build an outdoor delivery robot around visual technology. Liu Shaoshans low speed transport robot is currently known as the driverless car landing by a self driving start-up company. Liu Shaoshan believes that the market data is strongly attached to the scene, the port Park and the industrial park are two different scenarios, which must be trained according to a specific scene to achieve the desired effect. Why not drive a passenger car without a driver? Why not call an outdoor robot a driverless project? The capitals pursuit of unmanned driving is a reason, because the imagination of the market is too big and can even change the whole large logistics industry in China. It is important to know that the income of this industry can account for 20% of Chinas GDP, and the industry has a lot of pain at present. Liu Shaoshan first analyzed the reasons for the fever of self driving business. However, according to Liu Shaoshans judgment, the industrial chain of the car is too long, the driving scene is too complex, the targeted road infrastructure is not perfect, the real pilotless driving will take four or five years, and the passenger car market is a protracted war. Liu Shaoshan believes that unmanned driving will be a slow process, from a technical point of view of a long process, to verify the cost of a technology is also great, even if these are completed, and finally to see the popularity of the market. For the current unmanned venture, Liu Shaoshan believes that the unmanned venture company, which is the main driver in the car market, will have a longer landing cycle. If there is no other business, the final result may be bought by the big factory. There are advantages and disadvantages of the two solutions to the present driverless way. There are different opinions on which to adopt in the industry. Liu Shaoshan believes that in the passenger car market, the speed of the car decides that unmanned vehicles must be able to see 200-300 meters or even farther, so the laser radar is still necessary. Only in the last two years when everyone was driverless, most companies used a rotating mechanical radar, and many companies are doing it now. The scheme of the state radar. For low speed robotic products, the speed may be only 20-30 kilometers per hour, and Liu Shaoshan thinks it can solve these problems without looking so far. Liu Shaoshan told NetEase intelligence that his business called a carrier robot solution was afraid to be overestimated. In the absence of the scene, I was more afraid of being overestimated, driven by capital speculation, so that it would be more tired. Liu Shaoshans caution can be seen in the companys operation and business development. Perceptln has reached the stage after the A round in financing, but the total number of employees is only 30. Liu Shaoshan explained, we do not have large-scale expansion, because it is necessary to eat the scene, to determine the market is feasible, then can be promoted on a large scale. It is Liu Shaoshans constant emphasis in the interview. In the process of communicating with the NetEase, Liu Shaoshan, with a slight reflection, said that at the beginning of the business, we were overly pursuing the technology and believed that as long as the technology was good enough, we would be able to fall to the ground. But from a market perspective, many companies choose what technologies to use to meet the scenario according to the needs of the scene. As a technical person, it is difficult to really understand this thing. In the first two years when artificial intelligence was hot, the pursuit of technology for capital made it easy for start-up companies to raise capital. But since this year, capital has shifted from paying attention to AI technology to paying more attention to landing on the scene. The bet is more cautious, and the market will be more calm. Liu Shaoshan said. So, what scenes of AI technology have landed, what is the next market and scenario? Liu Shaoshan said that artificial intelligence technology has been popularized in the field of video, especially the security and monitoring market. This scene only needs to extract peoples information. The technology needs is simple and clear. The domestic Dahua and Hikvision have made a new security ecology based on the AI. We have entered the era of multimedia interaction, and any technology related to multimedia data understanding and analysis is valuable. Liu Shaoshan said. What is the next market and scenario? Liu Shaoshan said he was more optimistic about the combination of AI and medical care, because the pain in the market was large and demand was even greater. As long as we can identify the scene to solve the pain point, the start-up company will always have its own place to live. Note: exclusive interview with AI hero belongs to NetEase intelligent studio. It is an insight into technology trends with industry people to capture industry opportunities. Reprint please indicate the source. Click to see the topic of AI hero.