Unveiling the "Commander" behind the unmanned forklift army

As the mobile robot market becomes increasingly crowded, the competition between companies has evolved from product strength to comprehensive strength. As one of the key factors for the large-scale implementation of mobile robots, the importance of scheduling and control systems is self-evident.

Gaogong Mobile Robots noticed that according to the technical capabilities of previous mobile robot manufacturers, a site could dispatch dozens or hundreds of robots at the same time, and its dispatch system was considered good. However, as manufacturers' requirements for cost and efficiency become higher and higher, hundreds or thousands of AGVs will work together and operate independently, and the environment they face will change from static and single to dynamic and complex scenes.

How can mobile robot companies find the optimal solution in the shortest time and balance the contradiction between vehicle efficiency and quantity? This requires mobile robot companies to break through single-machine automation and achieve group intelligence. Undoubtedly, this method is extremely challenging for the dispatching system capabilities of manufacturers. Currently, there are very few robot companies that have commercialized and dispatched hundreds of AMRs in a single scene, and Muyi is one of them.

In order to explore the secret of how hundreds of unmanned forklifts of Muyi are dispatched in the same scene, this article interviewed Wu Wenlong, a senior product solution expert of Muyi, to reveal to us the difficulties of the dispatching system and the advantages of the Muyi dispatching system.

Muyi has been deeply involved in the logistics industry for several years. Based on its deep understanding of scenarios and the trend of evolution from low-speed single intelligence to high-speed group intelligence, Muyi adopts a road line scheduling design, which is different from the navigation scheduling used by traditional forklifts.

In 2020, Muyi Robotics implemented a large-scale cluster dispatching and control system solution for hundreds of units, opening up a new operating mode. Through the core dispatching system, it connects the goods, vehicles, and labor on site, from a single point to multiple sides, and finally forms a powerful data network, and then uses the data network to conduct digital management and operation of the site.

Specifically, Muyi's self-developed scheduling system adopts a distributed service architecture. The system can quickly access thousands of robots to prevent data congestion caused by large-scale data access. It has the characteristics of high performance, hierarchical storage function, consistency of multiple copies, disaster recovery and backup.

At the same time, based on data flow and modular system architecture design, Muyi's scheduling system can adapt to the rapid changes of future business, facilitate business decoupling, rapid expansion, convenient system upgrade by module, and quickly respond to customer needs. All modules are standardized and reused, which is conducive to the rapid and stable development of new businesses, organizational structure optimization, so that developers can focus on a certain business, convenient code management, database optimization, and in-depth research and development.

Forklift application scenarios are mostly large logistics parks or industrial parks, where the WiFi signal or 4G, 5G signal may be poor. In order to maximize the guarantee that the forklift is always online, the industrial WiFi, 4G and 5G networks of Muyi's AGV can be configured to automatically switch, ensuring high availability in any scenario.

In addition, the professional vehicle information platform can upload vehicle information to the cloud or customer terminals in real time, and provide complete data processing and analysis, finding ways to improve overall operational efficiency from the operational data center. Muyi uses forklifts as data collection terminals to provide customers with complete IOT solutions.

With the advantages of the Wood Ant scheduling system, Wood Ant has completed the batch implementation from indoors to outdoors.

For AGV or AMR, the premise for simultaneously dispatching hundreds of unmanned forklifts in complex scenarios is that the manufacturer's dispatching system must adapt to the customer's on-site environment. For example, how the manufacturer uses the dispatching system to allow the vehicle to arrive at the designated location at the correct time, complete the predetermined actions, and then complete the task in an efficient manner is extremely challenging for the manufacturer.

From the perspective of application environment, the scheduling requirements for indoor and outdoor are also different. From indoor to outdoor, the core difficulty that AGV needs to overcome lies in perception capability.

Generally speaking, the indoor environment is relatively simple, and it is less difficult for manufacturers to deploy AGVs. In this case, the focus of the dispatching system will be on the control of narrow channels and the traffic control of the vehicle itself.

Outdoors, the environment is more complex, and most of the scenes involve human-machine interaction, so the AGV has a higher speed requirement. At the same time, outdoor scenes also involve the interaction of traffic control equipment such as traffic lights. In addition, as the outdoor boundary expands, AGVs usually have to run several kilometers. When a fault occurs, human resources are required to go to the scene to handle the exception. Therefore, it is necessary to achieve a wider range and more accurate perception.

To this end, Muyi has specially added remote driving technology as an auxiliary means of outdoor dispatching. When the AGV encounters an obstacle or failure, the staff can control the forklift to intelligently avoid obstacles indoors, improving the safety of AGV outdoor technology.

Wu Wenlong pointed out that the so-called dispatching means that the robot, based on its own navigation and positioning, needs to be told by the dispatching system how to get from the starting point to the end point, while the question of "where to go" for the car is generally given directly by the business system, so "how to go" is the key to whether the dispatching system is intelligent or not. The standard for evaluating the quality of the dispatching system lies in how the vehicles in the entire system are reasonably allocated to complete the task in an efficient manner, that is, whether it is the global optimal.

To evaluate the global optimum, manufacturers need to have a deep understanding of customer business scenarios.

Wu Wenlong said: "To achieve the best overall situation, the business scenarios must be decomposed into technical indicators and then converted into scheduling algorithms. This is closely related to the company's strong R&D capabilities, and this is also the key to Muyi's commercialization of hundreds of forklifts in scenarios where humans and machines are mixed."

Taking the express delivery center as an example, the working environment attributes of the express delivery center are typical human-machine mixed type.

As we all know, the working environment of the transfer center is complex and has many obstacles. There are goods on both sides of the site, and the unmanned forklift has to pull the goods unloaded on one side of the site to the loading position on the other side of the site. But in fact, after more than 20 vehicles, the concurrency problems gradually increase. At this time, the R&D personnel must optimize the overall scheduling architecture. After more than 50 vehicles, the scheduling system needs to play a role in the predictability of vehicles to adapt to complex and changing scenarios.

When the dispatching system sends a command, there are many options for the robot's path. How to choose the best route to make the robot work more efficiently requires the dispatching system to add some preconditions. The implementation difficulty of the express delivery center scenario is higher than that of ordinary industrial warehousing, but it also better reflects the advantages of large-scale cluster dispatching.

In common e-commerce warehousing and industrial warehousing, the demand for AGV dispatching systems is relatively small. E-commerce warehousing rarely uses flat pallet handling scenarios, and generally only a few forklifts are used for regional selection from the warehouse to the store; while in industrial warehousing, the workpieces from the raw material warehouse to the production line are relatively large, and forklifts are needed for line-side transportation. However, in fact, the line-side material collection center does not require a high demand for forklifts, and only needs to transport a dozen or twenty pallets a day.

Therefore, compared with e-commerce warehousing or industrial warehousing, it is more difficult for the express delivery center to dispatch hundreds of AGVs. After the express delivery center unloads the goods, forklifts are needed to deliver the goods to the various departure directions, which will inevitably involve the operation of forklifts. At present, Muyi uses the express delivery hub center as its base and provides a full-stack handling solution through "group autonomous driving" technology. It has achieved the day and night operation of nearly 100 unmanned forklifts in a single warehouse. At the same time, it will spread across the country from point to surface, and carry out a "blooming" style of comprehensive and large-scale replication of scenarios.