Commercial delivery service robots, as a new species, have gradually entered people's lives in recent years. Unlike well-known industrial AGVs/AMRs, commercial delivery service robots focus on providing flexible, adaptable, and customized commercial delivery services. Their ultimate service recipients are individual C-end users, while the owners or providers of the equipment and services are the B-end clients of commercial delivery service robot companies, presenting a B2B2C business model. For delivery, high efficiency and low cost are the ultimate goals. Wheeled robots are currently a relatively mature robot product form suitable for commercial operation. UDI Technology, as a "veteran leading player" in the commercial delivery service robot field, also uses wheeled delivery service robots as its main product and service platform. Through continuous refinement and iteration of its technological solutions and large-scale product deployment, it continuously provides valuable services to both B-end clients and C-end users.
Currently, commercial delivery service robots can be categorized into three types based on their application scenarios: indoor delivery robots, outdoor delivery robots, and general-purpose indoor/outdoor delivery robots. Indoor robots represent the fastest-growing commercialization scenario, with widespread applications in hotels, restaurants, entertainment venues, shopping malls, and other commercial settings. UDI Technology's indoor robots have covered over 600 cities nationwide, partnered with more than 9,000 clients, and provide guided delivery services to over 250,000 people daily, with a cumulative service volume reaching 800 million people.
From a robotics architecture perspective, robots in any scenario will include sensing and perception modules, decision-making and planning modules, control and execution modules, and human-computer interaction and communication modules. This article will detail the construction of robot architectures in different scenarios from three dimensions: scenario openness, driving speed, and delivery distance.
Indoor delivery robots are used in enclosed environments with extremely low speeds, typically covering delivery distances within 0.1 km. The robot's chassis technology, a core technology in this scenario, is primarily reflected in its sensor architecture, electronic and electrical system architecture, and chassis mechanical architecture. The sensing and perception system employs a perception architecture based on single-line LiDAR, a vision camera, and an IMU. This technology maximizes the balance between cost control and performance advantages, and currently faces few major technical challenges. The focus is more on handling boundary situations that arise during extensive real-world operation while ensuring safety. As these boundary situation cases accumulate and are learned, the robot system will gain increasingly intelligent operational and delivery capabilities through efficient upgrades, thereby better providing value services to customers and users. The electronic and electrical system architecture, as the core computing power carrier of the robot system, has been upgraded by UDI Technology through independent research and development from a distributed topology architecture to a centralized bus topology architecture. It utilizes NVIDIA's early Jetson TK1 processor and later upgraded NANO processors to provide continuous computing power support for the robot system. As the core of a robot's obstacle-crossing performance, the chassis mechanical architecture has been upgraded by UDI Technology from its proprietary 6-wheel mid-mounted suspension differential chassis to a 4-wheel composite suspension chassis, improving the robot's adaptability and passability in harsh indoor environments.
Outdoor delivery robots encounter more complex scene subdivisions than indoor robots. UDI Technology has entered the field of low-speed delivery robots operating in closed or semi-closed environments, employing a sensor and perception system architecture primarily based on multi-line LiDAR, visual cameras, ultrasonic radar, IMU, and GPS. Even in closed or semi-closed environments, the robots operate in outdoor conditions, encountering more complex traffic elements such as pedestrians, motor vehicles, non-motorized vehicles, road maintenance, etc. Therefore, the sensors at the perception end need to acquire as much information as possible about the external world. Through data fusion algorithms, the robot can perceive and classify various surrounding elements. Highly secure multi-line LiDAR and cameras with richer perception information are also used as primary perception devices, similar to indoor robots. Meanwhile, outdoor robots operating in low-speed scenarios emphasize safety, and the related perception and planning control algorithms can be transferred from indoor robot strategies. This explains why UDI Technology was able to migrate from an indoor robot technology architecture to an outdoor robot technology architecture in a relatively short time and quickly achieve operational deployment. As a carrier of computing power, the electronic and electrical system architecture is more complex than the indoor perception system, generating a large amount of data throughput and processing requirements. A higher computing power electronic and electrical system is also a necessary condition. UDI Technology has designed an electronic system for outdoor low-speed robots that uses dual NVIDIA Jetson TX2 computing cores. In the future, it will also launch an electronic system using NVIDIA's latest Jetson ORIN. The use of automotive-grade components and design specifications also provides high reliability support for the electronic system of outdoor robots.
In addition to the applications in closed or semi-closed scenarios mentioned above, outdoor delivery robots also have low-speed applications in open road scenarios and high-speed applications in open scenarios. At the sensing and perception level, the two main international development directions at present are a pure vision-based solution represented by Tesla, and a multi-sensor fusion solution based on LiDAR, represented by Google's Waymo and most domestic autonomous driving companies, using multi-line LiDAR + vision camera + millimeter-wave radar + ultrasonic radar + IMU + GPS. The first solution pursues extreme cost-effectiveness but introduces safety risks, while the second solution ensures safety but is not cost-effective at present. The above solutions are all based on single-vehicle intelligence and single-vehicle autonomous driving. However, the dual-smart city (smart city infrastructure and intelligent connected vehicles) that my country is currently developing offers a different approach. By vigorously developing smart roads and other infrastructure based on C-V2X, strengthening roadside intelligence, and leveraging powerful 5G/6G communication network technology, richer perception information and redundancy compensation information can be provided for outdoor unmanned vehicles/robots. This will further improve the safety of single-vehicle intelligence while also increasing the operating efficiency of unmanned vehicles/robots, thereby providing better services for people's production and life.
In summary, from indoor delivery robots to outdoor open-road delivery robots, delivery robots are delivering over increasingly longer distances, serving wider areas, traveling at higher speeds, and facing increasingly stringent safety requirements. However, the essence of delivery robots remains the same: reducing simple, repetitive, and heavy human labor while providing people with efficient and low-cost value services. UDI Technology, adhering to its corporate vision of "Let robots shuttle through the streets and alleys of every city!", hopes to provide higher-quality robot delivery services to more people.
