/PRCWT/Nanjing, December 01, 2023 - From November 24 to 27, the 7th National Cluster Intelligence and Cooperative Control Conference and the first National Cluster Intelligence Technology Challenge hosted by the Chinese Command and Control Society in 2023 was held in Nanjing, the ancient capital of the Six Dynasties. More than 60 participating teams from outstanding universities and scientific research institutes in the country totalled more than 400 players. In the competition stage of more than 2 months, from online simulation competition to physical testing and on-site defense fierce competition, continuous innovation, detonating the industry technology development boom. The intelligent algorithm Cloud simulation platform (hereinafter referred to as "RflySim Cloud Platform") developed by Beijing Zhuoyi Intelligent Technology Co., LTD and its Feisi Laboratory provides a full range of intelligent simulation solutions, which has won the praise of more than 200 experts in the industry and all the participating students.
Policy-oriented industry technology trends are evident
This National Cluster Intelligent Technology Challenge (hereinafter referred to as the "Challenge") is the first time to be held. With the theme of "swarm express" and the goal of promoting technical exchanges in the field of cluster intelligence, the challenge focuses on the realization of key technical algorithms, requiring participating teams to use multiple quadrotor drones to achieve efficient cargo distribution under centralized command and dispatch control. It is divided into three parts: online simulation competition, physical test and on-site defense. After fierce competition,10 teams stood out.
Online simulation stage: The participating teams debug online and offline algorithms through the RflySim Cloud platform. During the competition, each team only needs to load its own algorithm to test the algorithm effect in specific cargo distribution task scenarios, which greatly improves the convenience and efficiency of the competition. In addition, the RflySim Cloud platform provides team management, simulation environment dynamic generation, ranking, test issue and other functions, and provides one-stop services for the participating teams such as pre-competition online training, simulation test issue, online real comparison test, ranking results review, record playback, etc., which stimulates the innovative thinking and technical ability of the participants.
Offline physical test and defense stage: Zhuoyi intelligent supports teams to carry out real flight tests, and feedbacks real flight videos and real flight data for them, in order to more intuitively understand the flight status and performance of the UAV, and constantly optimize the algorithm. During the meeting, the RflySim Cloud platform appeared on the site of the cluster control conference, showing the functions and advantages of the platform to industry experts and participants, which aroused great interest from everyone, and waves of visitors came to communicate and consult to find out. Experts and professors from Beihang University, Western University of Technology, Beijing Institute of Technology and many students from the winning teams of the competition all said that the technology of the RflySim Cloud platform is very advanced, and the use of the challenge also witnessed its advantages.
At the conference site, the person in charge of Zhuoyi Intelligent Simulation &AI Division introduced the unmanned intelligent system solution in detail for many industry research institutes and university personnel attending the conference. Zhuoyi intelligent online and offline efforts, not only provides an innovative cloud platform to meet the diverse needs of participants, but also creates more value and opportunities for education and research personnel in related fields, and wins the recognition and support of the vast number of industry users.
Ingenuity casting makes unmanned intelligent research easier
Founded in 2015, Beijing Zhuoyi Intelligence is committed to the development, simulation, testing and training of unmanned systems, and is the world's leading developer and manufacturer of intelligent unmanned system solutions. Feisi Lab, the brand of unmanned intelligence education and scientific research under Zhuoyi Intelligence, is mainly committed to the research and development of unmanned intelligence cluster and ontology teaching and research platform. The RflySim Cloud platform developed by Zhuoyi Intelligent Flix Lab provides exclusive intelligent simulation solution support for this cluster challenge.
RflySim The Cloud platform is a platform developed for unmanned platform cluster algorithm verification, large-scale game confrontation simulation, artificial intelligence model training and other cutting-edge research fields, supporting public cloud and private cloud deployment, and is a comprehensive platform integrating large-scale refined model simulation and online development, debugging, and training of multiple types of intelligent algorithms, solving the difficulties of development environment construction and hardware computing power There are some pain points, such as high search, lack of training data and lack of model support. Its advantages are mainly reflected in the following aspects:
· Large-scale cluster
The platform provides a unified visual interface, and uses Docker Swarm and NFS as unified operation and sharing services of service clusters to provide operating environment and data resources for all cluster nodes. Mysql and Redis clusters ensure the data persistence and security of the platform.
· Refined unmanned equipment model
In the RflySim Cloud platform modeling system, each module of unmanned equipment needs to be modeled with high precision and realized in the real-time simulation computer, and the control system software or hardware can also be connected to form a closed-loop simulation of software-in-the-loop, hardware-in-the-loop, and virtual-real combination simulation.
· Algorithm access is simple and convenient
The platform provides customized competition scenarios for participating teams, which can debug online and offline algorithms. At the same time, the platform provides online intelligent algorithm access SDK, built-in multiple programming language cores, including C, C++, Python, etc., which can support the intelligent algorithm access and training of upper-layer users, and access interfaces with agents and simulation controllers through algorithms Interaction with physical effect calculation module.
· Online training of intelligent algorithms
The platform provides reinforcement learning, deep learning algorithm debugging and training framework, visual setting and modification of reward functions, fast model parameter modification, multi-dimensional viewing of training effects.
· Support the combination of virtual and real
The platform supports both physical and virtual nodes. Physical nodes have high model accuracy, virtual nodes build a real-installed digital twin model,1:1 high-precision restoration of physical nodes, through reasonable allocation of the number of physical aircraft nodes and virtual aircraft nodes, to better achieve the verification effect of large-scale cluster simulation algorithm in the controllable cost range.
RflySim Cloud platform is widely used for algorithm verification, online competition, laboratory construction, virtual-real combination test, open cloud platform, etc. Up to now, it has provided technical services for hundreds of universities and scientific research institutes.
Zhuoyi Intelligent Feisi Lab provides many offline and online special training courses and cases throughout the year, and the relevant content has been published on platforms such as public accounts and Zhihu.
The future of unmanned intelligence education and research and development is promising
With the increasing maturity of UAV technology, its application in scientific research, education, military and civil scenarios (emergency rescue, inspection and inspection, logistics, public safety) is more and more extensive, which leads to a huge demand for relevant skills. As a leading enterprise in the industry, Zhuoyi Intelligence has been cooperating with domestic colleges and research institutions since its inception to continuously strengthen the training of highly skilled talents. In the future, Zhuoyi Feisi will continue to promote the "unmanned intelligence +" strategy, unite the national and global relevant unmanned intelligence research institutions and industries, create relevant industries and ecosystems, and make unmanned intelligence research simpler.
