Abstract: With the rapid development of the social economy, traditional enterprises are seeking transformation and upgrading. The rapid rise of artificial intelligence technology has made it possible for the service industry to realize transportation, operation and automation services.
Currently, in an environment of fierce market competition and increasingly personalized, diversified, and intelligent products, traditional enterprises urgently need to transform and upgrade. Therefore, while strengthening human resource cost management and control, enterprises also place higher demands on workers, requiring employees to be freed from simple and repetitive tasks to provide more advanced personalized services. Service robots, which have emerged as a result, are among the most practical and life-oriented types of intelligent robots , offering the possibility of replacing simple, repetitive, and tedious labor.
The "National Medium- and Long-Term Science and Technology Development Plan Outline (2006-2020)" clearly defines intelligent service robots as: "Intelligent service robots are intelligent equipment integrating multiple high technologies that provide necessary services to humans in unstructured environments." The "2018 China Robot Industry Development Report" shows that the Chinese service robot market size is expected to reach US$1.84 billion in 2018, a year-on-year increase of approximately 43.9%, higher than the global service robot market growth rate. As an emerging high-tech industry, the research and development of service robots' core technologies and products has promoted the development of core technologies in the national artificial intelligence industry. Their application in service industry scenarios will create a new service model, alleviate the shortage of talent in the service industry, improve service quality, and promote the intelligent upgrading of the industry.
Service robots are essentially a fusion and implementation of multiple technologies, including voice interaction, navigation and positioning, motion control, back-end scheduling and management, multi-sensor technology, and communication. All these technologies are indispensable for successful commercialization. To achieve precise robot service, environmental recognition is required in conjunction with location information. This necessitates the use of SLAM (Simultaneous Localization and Mapping) technology. SLAM means simultaneous localization and mapping, where the robot starts from an unknown location in an unknown environment, performs self-localization based on position estimation and sensor data during movement, and simultaneously builds an incremental map. LiDAR is a crucial entry point for SLAM. The principle of LiDAR is to use triangulation or Time-of-Flight (TOF) to measure the distance between itself and surrounding objects by emitting n laser beams, obtaining highly accurate distance information. Compared to visual positioning, the SLAM + LiDAR solution offers significant advantages. It allows the robot to maintain a positioning accuracy of 0.01-0.1 meters while fulfilling service commands, analyze, judge, and select the surrounding environment, obtain an accurate map, plan a path, and ensure safe movement from one location to another without collisions.
From an appearance perspective, service robots move in two ways: gait and non-gait. Gaited walking includes hydraulic and motor control; while these robots are more humanoid in appearance, they suffer from poor stability and slow movement. Non-gait walking mainly consists of longitudinal and lateral control. These robots can adopt different control strategies based on speed during movement and obstacle avoidance to maintain overall stability. Voice interaction typically involves receiving information from external input systems, usually voice recording. After voice decoding, the information is imported into a pre-defined knowledge base for semantic matching, logical processing, and then voice synthesis. The output is then selected as either voice or text, depending on the requirements. The most important parts of the entire module are speech recognition and semantic analysis. In the early 21st century, due to increased computing power, speech technology made significant breakthroughs. Semantic analysis and understanding require breakthroughs in deep learning computing. Image and facial recognition technologies complement the development of voice and semantic interaction. With the help of image and facial recognition technologies, information can be retrieved more quickly through image search, and new interaction methods can be created, allowing robots to operate more intelligently.
UDI Technology has accumulated extensive experience in practical applications. With robot walking technology built on the framework of autonomous driving technology as its core, and rich experience in embedded R&D, it has independently developed technologies such as SLAM/VSLAM positioning and navigation modules, robot chassis and CUDA high-performance computing platform. It is committed to promoting the commercial application of low-speed autonomous driving technology in the last three kilometers of last-mile delivery and providing stable and reliable service robot mobile platform solutions for global customers.
Our independently developed robot chassis (as shown in Figure 1) possesses functions such as autonomous path planning, walking control, autonomous obstacle avoidance, autonomous charging, and elevator control. The chassis integrates high-precision motion control components and sonar collision sensors, providing the robot with accurate and reliable walking control and balance capabilities; it can bear a weight of 100kg and walk on slopes up to 10 degrees, adopting different control strategies according to different speeds to maintain overall stability.
Figure 1 Self-developed robot chassis
Our service robot, Youxiaomei (as shown in Figure 2), is based on indoor autonomous driving technology and utilizes multi-sensor fusion technology, integrating sensors such as vision, LiDAR, and sonar to accurately detect the surrounding environment. It can autonomously plan paths, avoid obstacles, and flexibly complete tasks such as intelligent guidance and transportation. Currently, Youxiaomei is widely used in commercial venues such as shopping malls, hotels, restaurants, and KTVs, providing services such as greeting and delivery. At the front desk, Youxiaomei identifies guests using facial recognition technology and greets them. After guests check in, Youxiaomei uses its proprietary vision and LiDAR algorithms to efficiently and accurately guide them, providing voice prompts when leading them to restrooms or specific locations. The high-definition touchscreen displays customized content and human-computer interaction interfaces; by entering the relevant password, guests can open the storage compartment and autonomously complete tasks such as delivering food and drinks. This detailed experience greatly improves guest satisfaction.
Figure 2. Service robot Youxiaomei
Our newly developed outdoor delivery robot (as shown in Figure 3) utilizes multi-sensor SLAM global positioning and navigation technology, perception technology, and machine vision to achieve its walking technology. It uses AI algorithms to predict pedestrian routes and manages the operation from a cloud-based backend. Its primary applications include food delivery within a 3-kilometer radius of closed campuses and schools, as well as delivery to nursing homes and villa areas. The robot's four storage bins interface with food delivery or supermarket systems. Users can open the bins by scanning a code, place their items inside, and receive a text message notification upon delivery to the designated location. Users can then retrieve their items by entering the pickup code received in the text message. The outdoor robot has completed basic and road testing and will be commercially available within the year.
Figure 3 Outdoor delivery robot
According to current market data on Youdi's service robots, the average time for a single delivery/delivery is 3 minutes; the walking speed is 1 m/s, equivalent to a normal human walking speed; a single robot can work for 7-12 hours on a full charge; it can travel 10-20 kilometers per day, with a peak travel distance of 28.8 kilometers. In the future, Youdi Technology will invest more resources in the commercial technology research and development of service robots, reducing labor costs for businesses and allowing employees more time and energy to provide customers with humanistic care and personalized services.
