Location: Home >Blogs> Universal and voice robot Rokid cooperate, what are they going to do together?

Universal and voice robot Rokid cooperate, what are they going to do together?

Asian Industry Network News: What will the car of the future look like? When you mention this question, you may think of driverless cars, you may think of all kinds of exaggerated concept cars, or even spaceships. But the answer to this question is not so unrealistic, it is actually very simple: in the future, cars will become part of artificial intelligence.

It can be said that in the future, the car itself will become a kind of robot. And AI will become the core that drives these Robots. When you’re ready to go out, just make a call to the robot at home, and he can let your car drive to your door through self-driving. After you get in the car, the artificial intelligence in the car will automatically help you adjust the seat to the most comfortable position and adjust the air conditioner to the most suitable temperature.

What’s more, it is able to communicate with humans directly through speech. Tells you the way forward, narrates the surrounding scenery for you, reminds you of impending danger. All of this will make you feel like you are no longer sitting in a cold Industrial machine, but traveling with your best friend.

In order to bring such a scenario into reality earlier, more and more auto giants have begun to join forces with technology companies. Following the establishment of a joint venture between Volkswagen and Mobvoi to explore the application of artificial intelligence in the automotive field, Buick, a subsidiary of General Motors, and Rokid, an AI company, have also reached a cooperation. Combined, an artificial intelligence home platform is built.

Their results send a signal: the era of artificial intelligence is really not far from us

The platform is called “VELITE”, and it is a set of intelligent buildings that you can only see in science fiction movies. This time, the two sides jointly moved to the Shanghai Bund.

The two-story architectural design, large glass windows, the scenery of the Bund, plus a Buick extended-range hybrid concept car VELITE 5… All these remind me of Tony Stark in the movie “Iron Man”. Super cool beach house.

Speaking of Iron Man, you will definitely think of Jarvis, the robot assistant in his home. Jarvis in the movie is the omnipotent patron saint around Stark, and he has become the best interpretation of “robot” for many technology fans.

In Buick’s platform, it is the intelligent voice robot from Rokid who plays “Jarvis”.

Like in the movie, the Rokid intelligent voice robot also acts as the steward of the entire platform, taking control of various smart devices in the home with the assistance of its partner Lifesmart. This includes smart air conditioners, fresh air systems, smart door locks, audio appliances, and more. As long as you call Ruoqi, she can complete the task.

In addition, Rokid, like Jarvis, can communicate directly with users.

When you drive back home, just say “VELITE! I’m back!” to the Rokid voice robot when you enter the door, and she will automatically turn on the lights, air conditioning and fresh air system in the house for you. And also play your favorite music. How “warm” such a scene is!

You may ask: This artificial intelligence butler has a bird connection with the car? But in fact, Rokid’s set of voice interaction logic has many similarities with many usage scenarios of in-car HMI. When we use voice in the car, it will also involve scenes such as playing music and adjusting the temperature of the air conditioner. Imagine that with the help of autonomous driving, the car itself will also act as an intelligent tool to connect with the artificial intelligence platform in the future.

In fact, whether it is a Jarvis-style intelligent robot or in-car interaction, the development direction is to make the machine more humanized. It can make users feel that they are not communicating with a cold machine, but interacting with a friend who understands them.

1 2 3 Next > page

A local AI company founded by a “hermit”

For the automotive industry, Rokid is definitely a “little fresh meat”. They are an AI technology company with R&D centers in Silicon Valley, Beijing, and Hangzhou, established in 2014. I’ve never done anything in the auto industry before.

The founder, Misa, worked in Ali before and is the head of its most mysterious black technology department: M Studio. One of the original important research directions is deep learning.

Misa himself is a “hermit”, he is very low-key, rarely gives interviews, and rarely tells stories in public.

But those who know him will tell you. In terms of product development, the team led by Misa is absolutely paranoid. The Rokid team often “quarrels” over product development issues. Rokid has a strict pursuit of experience, and often overturns the previous plan due to a subtle experience and starts over.

What kind of team can be so paranoid? Here’s a picture to take you to know this team of great people:

Many of these fields that you have seen may be things that we have never heard of, but Rokid’s team is full of scientists in this field. In these fields, the number of people who can read a doctorate in a top college may be counted by one hand every year. However, they were attracted by Rokid’s vision and joined Rokid to become a strong backing to ensure the advanced technology of Rokid. Their research interests include specific types of voice recognition, timbre variation, multilingual ASR, emotional speech synthesis, neural network modeling, deep learning, semantic understanding, and more. The purpose of all these researches is to allow machines to learn new knowledge and express them with emotion, rhythm and rhythm like humans.

In 2016, Rokid released the company’s first mass-produced product “Ruoqi Alien”. This is a cute artificial intelligence “girl” that looks like an alien. Her face is a projection panel that can Display various expressions and content. This unique Display structure is Rokid’s own patent. In order to mass-produce it, Rokid searched for suppliers all over the country, and finally found a supplier of iPhone flash cover in Japan to provide this panel.

Just like Jarvis, “Ruoqi” can communicate directly with users through voice. In addition to chatting with her, users can also ask her to help play music, check information, and control home appliances and lights through the partner’s API.

“Ruoqi, play a song for me.” “Ruoqi, what’s the weather like today?” “Ruoqi, I want to hear the news.” “Ruoqi, help me translate.” “Ruoqi, what is 250 times 360? ?” These instructions are the daily routine of Ruoqi users, and Ruoqi can deal with them calmly.

The Rokid team hopes that through products like “Ruoqi”, users can experience the future life, and the distance between people and artificial intelligence can be brought closer, so that artificial intelligence can be better integrated into users’ lives.

Ruoqi’s perfection is not achieved overnight. In order to make “Ruoqi” more humane, Rokid’s team has made great efforts to polish the product. For example, the most common device wake-up scenario in voice interaction. Like “Hey! Siri” on Apple devices and “Alexa” on Amazon Echo, the wake word they use is at least three syllables. In Chinese language logic, two-syllable words are generally used to greet each other, because the three-syllable appellation will appear too serious or unnatural, even if the other person’s name is three characters. People also choose to call them two-letter nicknames.


So during the research and development process, the Rokid team decided to remove “hey” and use only the two-syllable “Ruoqi” as the wake-up word. Such a requirement brings great challenges to product development, because two-syllable wake-up words will bring about a very large false trigger rate. The user may wake up the robot by saying “I’m going”. The team is determined to do it, even at the risk of delaying the product launch. After constant algorithm updates, Rokid’s current products are rarely awakened by mistake. In addition, the Rokid team also hopes that the voice expression of the robot itself is like a real person, becoming a member of the family. To this end, they found 200 dubbing teachers, and finally found the right voice. Coupled with the algorithm developed by themselves and the tuning by a music doctor with absolute pitch, they finally formed “Ruoqi” today with rhythm, rhythm, Emotional tone.

In addition to the voice interaction mentioned above, Rokid has many “humanized” advantages, such as the user preference system: through the accumulation and analysis of user data, Rokid can continuously learn and deepen its understanding of users. For example, when the user asks to play music, Rokid will play the user’s favorite music style.

In fact, the core technology behind all these “humanized” experiences is the Machine Learning Algorithm in the field of AI. The feedback generated by Ruoqi in the process of interacting with users is the result of the algorithm’s analysis of a large number of users.

GM and Rokid may have further plans

As we said at the beginning, car companies are paying more and more attention to AI technology.

But you must not think that any technology start-up company will be able to pull the car company to do technical endorsement in the future. In the eyes of car companies, most start-up technology companies play with pediatric technology, and they don’t even have any value in talking about cooperation. Because compared with smart hardware, the automotive supply chain is too complicated. Looking at what Lao Luo was abused by the mobile phone supply chain, the car is tens of thousands of times more complicated than the mobile phone supply chain.

Therefore, the technology companies that car companies look at must have the ability to make breakthrough innovations. GM and Rokid have joined hands to not only make cars more “human” through intelligent means, but more importantly, Rokid has given cars a huge blueprint for the future world. Through this blueprint, GM can see from Rokid the various possibilities of artificial intelligence in cars in the future, among which the following points are particularly important:

Improve product interaction experience: In the current in-car interaction scenarios, both voice interaction and user preferences, which are deeply cultivated by Rokid, have important application space. Voice has become one of the most suitable interaction methods to replace physical buttons or touch screens in the industry. The optimization of user preferences allows users to obtain a better experience when using functions such as in-car navigation, multimedia entertainment, and LBS. In the above-mentioned scenarios, although traditional car companies have accumulated related technologies, they often do not do enough in localization and humanization, and do not understand the interaction habits of Chinese users. As a local technology company that focuses on interaction, Rokid can provide many optimizations and suggestions for GM in this regard.

Gather talents in the field of AI: As we introduced in the previous article, AI is the core technology behind human-computer interaction in the future. In addition, in autonomous driving, AI technology is also a top priority in image recognition and driving strategies. In the automotive industry, AI is one of the areas where breakthrough innovation is most needed. However, innovation requires talent. At present, there are not many outstanding talents related to AI in the industry, and most of them are from technology companies. The Rokid team is one of them. Rokid also has laboratories in Beijing and Silicon Valley, and is one of the few startups with two laboratories. Therefore, through this cooperation, GM can obtain a reliable partner in the field of AI.

The two sides have similar tempers: the car company’s style is safety first, and even if a new technology has a little safety risk, it cannot be implemented in the car scene (because any loophole may have fatal consequences). The R&D logic of technology companies that make products first and then rely on user feedback to iterate is simply unthinkable in the automotive industry. In the eyes of tech companies, the cautiousness of car companies is typical of old stubbornness. So all along, the two industries have been full of conflict in style: the automotive world is full of awe and the tech world is always full of climaxes. But unlike ordinary technology companies, Rokid’s style is low-key, rigorous, and solid, and in terms of AI technology, Rokid has always adhered to the goal of “Serious AI”. Rokid hopes to truly guide AI technology into the life of service users, rather than just being a fancy smart hardware that sells well. This kind of behavior may be unique in the technology circle, but it is really pleasing to the eye in the automotive industry.

To tell the truth, although the content released by the two parties this time is only for some concepts of future life. But through the above analysis, we can boldly predict that GM and Rokid may cooperate on mass production-level interactive technology, especially for the Chinese market. This may also be the reason why GM chose Buick, a brand with strong localization. It seems that as auto giants pay more and more attention to China’s local AI technology research and development strength, cooperation with auto brands is no longer the patent of international technology giants such as Nvidia and Intel. Technology companies like Rokid, which are deeply involved in a certain field, will usher in their own opportunities.

The Links:   3BSE018172R1 3BSE013234R1