In the public perception, the voice of connected cars has been low, and the wave of publicity that began in 2014 is long gone. But in the eyes of IT companies, the show has just begun. Hype concept is the appearance, grab the territory is instinctive, go the right way is the key.
IT vendors and hosts have not yet had time to educate potential customers, so much so that most of the public still can't distinguish between Telematics and Telematics. But it doesn't matter, what customers need is intelligent systems to do as much as possible, while they just have to enjoy life. The more trouble the customer is put through, the less likely the product is to succeed.
BAT, which is the biggest player in the Internet and mobile Internet stage, has already occupied the biggest ecological position. All other APPs want to start a new stove to do platform-level products, has been nearly impossible. Do not take into account the BAT's own strong R & D strength, they only have to "buy buy buy" will be able to deal with almost all the startups, the latter always have to face the cash to return to the obedience as well as with the BAT arm wrestling the dilemma of choice.
In the era of connected cars, the situation is a bit different. Although huge such as BAT is still keen to release "strategy", "platform", but startups see new opportunities, without necessarily looking up to the former nose.
In the three-layer structure of the Telematics industry chain, TSP (Telematics Service Provider) occupies the core position of the industry chain. TSP is connected to automobiles, in-vehicle equipment manufacturers, network operators, and content providers, and thus has become a role that hosts, telecom carriers, and IT vendors are striving for. In addition to the platform, BAT chose not to personally do the product, and herein lies the reason.
The Internet of Cars has not yet been perfected, giving startups more opportunities.
Bottek is an active startup. In the past two years, it has cooperated with Baidu, United Electronics, Huawei, NXP and half of the companies in the car networking chain. From this all the way through can be seen, although Botai treat cross-border cooperation most actively, but the cooperation with car companies are "point to point", it is difficult to get a strategic cooperation opportunities. Because of the pure software company, in the car networking layout is bound to not be able to take into account.
Speech knowledge of the entrance to the war
Speech interaction at the beginning of Google (Google Now), Microsoft and Baidu search engines in doing, with the intention of opening up the fences of data in different languages, the global users to achieve the whole network, the whole data coverage, and seek greater advertising interests. But soon, the whole chain of car networking are aware of the huge application value of voice interaction in the car machine.
Driving is not convenient for touch operation, voice becomes the input control method. However, voice interaction is considered to have the potential to become the in-vehicle network, inter-vehicle network and in-vehicle Internet **** the same "entrance", since it is the soldiers must fight, resources quickly gathered in the head of the company.
The voice interaction platform for automotive applications has entered a white-hot stage. The number of products to reach double digits, they all do mobile APP, hoping to influence the user's habits. Moreover, our young users, first have a cell phone, and then have a car. It is inevitable to start cultivating customers from cell phones.
A few years ago the correct rate of speech recognition was not as good as it should be, with the development of artificial intelligence technology, natural language processing (NLP) and natural language understanding (NLU) engines are becoming more and more mature, and each claimed recognition rate is higher than 95%. Even so, customers always grumble about the unreliable 5%. The complexity of language performance in different contexts requires constant "feeds" of data to train the system and enhance its self-learning capabilities. The competition for speech recognition has transitioned to a competition for supercomputing and big data processing power, and the first-mover advantage of startups has been weakened.
In 2014, Amazon's smart speaker Echo, based on the Alexa platform, went live and opened up third-party development interfaces, spawning thousands of applications. As a smart home product, it easily migrated to the in-car environment. It's the first mature smart voice product.
Apple's HomePod came late and was not released until February this year, and was jokingly called "the bottom of the box". Relying on Siri as the representative of the intelligent voice assistant, Apple is aiming at the Internet of Things (IoT) era of intelligent interaction port, car networking is just one of the application scenarios.
During this period, the U.S. Nuance, domestic KU Xunfei, Sogou voice, cloud sound, Spectrum and other products have layout intelligent voice products. In last year's Ali and SAIC cooperation on the first Internet car, the use of Nuance voice interaction system; KU Xunfei just yesterday with Qichen signed a contract, earlier also signed a series of cooperation agreements with Chery, Changan, BAIC; out of the door to ask the Volkswagen China to set up a joint venture.
The BAT is no longer the way to go for startups to get their products off the ground. It seems that BAT is no longer dominating the entrance of TSPs everywhere on the Internet. However, BAT is still attempting to circle its own turf through capital operations.
Voice is just a tool, what customers really need is an intelligent assistant. The assistant can't just search and push a bunch of messages, but has to make decisions for the customer, or at least narrow down the options to a very small range. From this point of view, Siri is not a "competent assistant".
With the benefit of voice recognition, Tech Data Xunfei has released a smart car machine called "Flying Fish Assistant," which claims to support real-time demand changes, interruptions, and sleep wakeups, making it look like it has improved its intelligence and can hold all the conversations of everyday city life.
If an intelligent assistant is supported by a car company, it can control some non-safety items by voice: air-conditioning, stereo, phone, window up/down, etc. Even if it opens up non-critical controls, it will not be able to control the car by voice. Even when opening up non-critical controls, car companies are hesitant to take small steps. So much so that voice recognition companies have done little to build on this.
Tencent, which started a little later, is working with its peers in the field of voice interaction, but the latter does not yet have the strength to challenge KDDI. Tencent AI in car ecosystem, launched in the Bo Yue conference in the trend, but the Tencent statement does not do the operating system, still seeking to migrate social, do super APP.
Get the car end data, still have to be BAT
This time, BAT can come forward. Last year, Baidu World Congress launched the "Xiaodu car system", is a typical AI interaction system. "Xiaodu" built-in functions, intelligent voice assistant, face recognition, fatigue monitoring, AR navigation, HMI, car home connectivity, intelligent security. Xiaodu is more like the *** same entrance to the products of Baidu's various business groups.
Ali, on the other hand, offered its Zebra Smart Company and AliOS system, emphasizing that the latter is a cloud-based system. It can be updated like a cell phone system like OTA, but also like a car to achieve a long cycle life. The secret lies in the fact that with the help of computing and storage capacity on the cloud greatly reduces the pressure on the car side and extends the life cycle of the hardware on the car side.
The problem is that intelligent scenarios such as making decisions for users and controlling non-security devices in the car need to be assisted by car companies. In this way, whoever can get the data from the car company can smoothly access the TSP. thus the curve occupies the TSP port.
Zebra and SAIC reached an agreement, Zebra does not seek to obtain the underlying data of the car, but the user in the use of AliOS car generated big data, will be uploaded to Ali server. The driving map source data is deployed in the car and the cloud at both ends, and AutoNavi has access to "interactive information" in the background.
Ali has stronger negotiating leverage, and SAIC has concluded a strategic agreement, which is not possible with the size of the startup. Moreover, it has torn a hole in the data packages that car companies consider to be off-limits. Tencent's strong social advantage to implement the "offset strategy".
It is worth mentioning that Ali intends to use this as a model to talk about data cooperation with car companies one after another. For new energy startups, BAT intends to "ask" for a higher price to get all the data. But there is not yet a car startups into the game.
The essence of Telematics is to abandon the cell phone mapping of the car. The realization of intelligent scenes must be issued directly from the car network, the phone is no longer as a process node. This is not only more seamless, but also can cultivate the user to the car networking era, and then turn to the era of the Internet of Everything, when the cell phone will come to an end.
Tencent is the premier social service provider, more hope that the car connected products AI in Car, Tencent's social network to the car, in order to highlight their own advantages. However, Tencent's AI in Car, in addition to pointing to its own expertise in the content, perhaps should build a more pure TSP architecture, to develop a more needed decision-making assistant function in the car environment, rather than simply move WeChat on the car, at the same time, to reach a closer cooperation with the car companies, trying to let the latter in the series of products on the adoption of Tencent's front-end platform.
Investing in AI is investing in Telematics
How to kill the cell phone?BAT has come up with its own platform-level products. Baidu chose Carlife, a cell phone interconnection program similar to Apple's Carplay; Alibaba chose the Native App route based on Yun OS; with its huge user base, Tencent hopes to divide applications into channels to control the user experience and move towards the road of Super App.
Baidu's layout is earlier and more comprehensive, investing heavily in AI, trying to hold up DuerOS and Apollo two open platforms, the latter to all automotive developers to cooperate. Baidu claims to have more than 70 partners, but first-tier car companies, although most of them are on the list, have coincidentally avoided in-depth cooperation. Despite Baidu's determination to make a universal operating system + development platform for the car network, and also invested a lot of publicity sound, but as of now, no breakthroughs have been made. This is not directly related to Baidu's AI progress, car companies hesitate more because of the dominant right to doubt.
AliOS, developed by Ali, comes with its own platform attributes and cooperates with SAIC, Changan and Shenlong on AI, applications, social and information security.
Whether it is the super entrance of voice recognition, or the platform, or the data exchange between the cloud and the car side, all rely on the development of AI technology. the algorithmic training of AI provides a path for the car machine intelligence, scenario-based recommended services more humane. The more you understand the user's car networking, the more you will be favored and relied on by the user.
Cloud computing is the basis of AI arithmetic. Ali focuses on allowing car owners to get more intelligent services, while Baidu Apollo focuses on the field of automatic driving, so that the car is more intelligent on the road; Baidu DuerOS is to make voice interaction in the car more intelligent.
In the field of AI, Ali and Baidu will be in direct competition. Google will also become an important competitor if it can return to China. For Tencent, the dual drive of content and TSP is more competitive than the heavy content and light platform approach.
It is worth mentioning that last year, the Ministry of Science and Technology announced the "national new generation of artificial intelligence open innovation platform", Baidu's intelligent driving platform, Ali Cloud's smart city platform, KU Xunfei's intelligent voice platform, are directly related to the Internet of cars, only Tencent to obtain a medical imaging AI platform, and the previous three Out of step with the first three. It reveals that Tencent in the direction of AI strategy (light assets + heavy investment), there is a need to adjust.
With the development of AI, human-computer interaction will inevitably tend to simplify (reduce the cost of user learning to use). Vehicle-to-person, vehicle-to-vehicle, vehicle-to-various services connection relationship will be driven by AI. The super entrance status of intelligent voice will be consolidated to a greater extent.
The global automotive industry now totals roughly $2 trillion per year, while the future market for personal mobility, combining connected cars with autonomous driving and new energy technologies, will reach the $10 trillion level. Although the cake will continue to get bigger, but the future of Telematics products and strategy competition, as in the Internet era, whether to oligopoly situation, depending on who can control the "entrance" and "path".
Source: Tencent Motors