On the 100-day countdown to the 2022 Beijing Winter Olympics, the media who come to the "Ice Cube" for interviews and reports will be attracted by the intelligent robots that shuttle around in front of them. They have different shapes and have outstanding "working" capabilities. There are cleaning robots, disinfection robots and 5G food delivery robots. It can be said that their appearance not only facilitates daily work, but also becomes a beautiful landscape.
Among them, the most familiar robot is the mobile temperature measurement and epidemic prevention supervision robot "Benben". It can actively find people and measure their body temperature according to the prescribed route. Report to management. For those who do not wear masks in the area, he will also go to remind them to wear masks.
In addition, the 5G food delivery robot is also a highlight of this Winter Olympics. The robot adopts 5G indoor multi-functional cloud technology, integrates 3D semantic maps, and combines multi-modal autonomous navigation and obstacle avoidance intelligent walking capabilities such as laser radar, camera, and infrared sensors. Smart food delivery service. This "contact-free" catering delivery service can help the service guarantee during the normalization of epidemic prevention and control.
These robots working in the "Ice Cube" are generally classified as service robots. According to the China Robot Industry Development Report, the global robot market is expected to reach US$33.58 billion in 2021. As the epidemic is brought under control around the world, the robot market will gradually pick up. It is estimated that by 2023, the global robot market will exceed US$47.7 billion.
Although the prospects for the development of robots are good, they are also facing a lot of challenges. Now robots are facing problems such as incomprehension, incomprehension, and irrelevant answers. As the world's leading artificial intelligence data service provider, Datatang has been deeply involved in the field of artificial intelligence data services for more than ten years, has a deep understanding of the data needs of multiple business scenarios, and can help robots solve these problems through high-quality data solutions.
Speech Recognition Data Solutions
In a noisy environment, the speech recognition system of the robot will be disturbed, and factors such as accent and foreign language will also affect the accuracy of speech recognition.
In response to the above problems, Datatang can provide finished voice data covering multiple languages, channels, environments, and types, which can effectively improve the accuracy of customer voice recognition models. According to the individual needs of customers, Datatang can also use mature tools and platforms to customize data for customers, and solve the problem of "unintelligible" by robots in one stop.
Speech Synthesis Data Solution
Today's robots synthesize speech by collecting voice data of different emotions of the same person for learning. Due to the limited data scale, the synthesis effect is mediocre, and the synthesis results have poor emotional expressiveness, stereotyped answers, and cannot be highly anthropomorphic.
Datatang can provide speech synthesis data solutions with multiple pronunciation styles, multiple application scenarios, and multiple languages. Abundant sample audio resources can match the needs of sound characteristics in finance, securities, weather and other fields, and can quickly improve the anthropomorphic degree of speech synthesis.
Semantic Understanding Data Solutions
Due to the lack of high-quality training data, robots cannot fully understand the real intentions behind people's questions, causing the questions people ask and the robots' answers to be completely different.
Datatang's semantic understanding data solution supports multiple natural language understanding data processing requirements such as Chinese and English, traditional and simplified word segmentation, part-of-speech tagging, and syntactic analysis, which can quickly improve the ability of robots to understand semantics.
Face recognition data solution
In order to perform special tasks, many robots are equipped with facial recognition systems. Take the mobile temperature measurement and anti-epidemic monitoring robot "Benben" in Ice Cube as an example. Due to the cover of masks and hats, the number of facial biometrics available for identification is reduced. There are many types of cover and the cover degree is different, which seriously affects In order to ensure the accuracy of recognition, how to use the information of the unoccluded part for face recognition has become the key to the problem.
Datatang can provide high-quality facial occlusion and multi-pose face recognition data. By learning these data, it can improve the accuracy of "recognition of people without masks, and help the service guarantee during the normalization of epidemic prevention and control.
At present, most countries in the world are developing intelligent robot technology. Da Vinci surgical robots have opened a new era of global medical care. Military robots have become the key military security strategies of various countries in the 21st century. Industrial robots have shown their talents in various industrial fields. In addition, there are The sudden emergence of drones, robots used to accompany the elderly, household cleaning robots, etc., the army of robots is entering our lives.
With the continuous maturity of technology and market, the domestic robot industry will usher in broader opportunities. The combination of high-quality data and algorithms will also train robots with more thoughts and attitudes.