Hello everyone, I'm the person in charge of subject three, Zhuo Qing. On behalf of the third topic, the participants in the research on group intelligent collaborative operation and cognitive computing technology will report the relevant content of the third topic to the experts. The main participants in the third project include: Tsinghua University, Heilongjiang Dewo Technology Development Co., Ltd.
The content of the subject will be reported in the following five parts.
Firstly, the background of the project and the main research contents of the subject are introduced.
The project "Swarm Intelligent Autonomous Operation Smart Farm" is based on the current situation of my country's agriculture and implements the strategy of precision agriculture. Aiming at the many difficulties of the modern agricultural industry, applying a new generation of artificial intelligence technology to meet the needs of the upgrading of the agricultural industry.
The research content of group intelligent collaborative operation and cognitive computing of topic 3 includes three technologies: 1) Based on the characteristics of agricultural operations, that is, autonomous decision-making and task assignment of agricultural machinery, and human-machine integration technology under the non-structural and complex farmland environment; 2) High efficiency Reliable on-site wireless real-time communication and computing platform technology combining embedded edge and cloud; 3) Multiple and multiple sets of agricultural machinery autonomous driving and formation efficient collaborative operation technology in multiple links of farming, management and harvesting;
Combined with the specific production crop varieties of smart farms: the actual solutions proposed in the production of wheat, corn and rice: (1) autonomous decision-making and task allocation in swarm intelligence; formation motion control based on reinforcement learning; (2) human-machine based on intention recognition Achieve technological breakthroughs in collaboration and other aspects; develop an embedded computing platform that satisfies project demonstration and verification, and realize on-site real-time wireless networking and intelligent decision-making planning algorithms; (3) Finally, the efficient collaborative operation and production of agricultural machinery groups in all aspects of farming, management and harvesting are realized. The whole process is intelligent;
The local shows the relationship between topic 3 and other topics, as well as the internal main research content structure. Task 4 assigns cluster tasks to Task 3; Task 3 completes the internal autonomous decision-making algorithm on the self-developed real-time communication and edge computing platform, realizes field formation control, and realizes real-time feedback between humans and machines through the human-machine collaborative interface. The specific real-time instructions are sent to the agricultural machinery interface of the second task. Cooperate with the whole to complete the compilation of industry standards for agricultural intelligence, autonomy and unmanned operation.
(2) The main technical route and technological innovation.
In terms of autonomous decision-making and task allocation of agricultural machinery groups, according to the complex dynamic environment of farmland operations, the knowledge base established by other topics of the project is used to design and develop autonomous decision-making and task allocation algorithms. It mainly uses evolutionary game-based algorithm, bionic cluster decision-making algorithm and swarm intelligence optimization algorithm to realize group task assignment and scheduling.
In the formation and collaborative operation of agricultural machinery groups. On the one hand, the traveling characteristics of different agricultural machinery are very different, and on the other hand, the farmland has certain uncertain and unstructured characteristics. The reinforcement learning method is used to realize the optimization of formation control parameters and strategies, and to carry out real-time control.
Human-computer interaction is an important guarantee for the completion of each stage of the intelligent field operation of agricultural machinery groups, which is known, controllable and analyzable. Determine the boundary conditions for the safety and reliability of unmanned systems. On the one hand, operation monitoring personnel can understand the operation status of agricultural machinery through hand-held or worn electronic equipment;
Aiming at the task and environment of farm implements on site, a hardware platform capable of on-site wireless networking and embedded computing is developed. Complete the calculation of autonomous planning tasks, issue real-time control instructions to the control unit of the lower computer of agricultural machinery, and complete the collection of job information and operating status.
The main innovation points of the project include two aspects: 1) Propose a method for collaborative operation planning of heterogeneous agricultural machinery and swarm intelligence based on the dual drive of big data and knowledge graph, combined with artificial intelligence methods such as bionics and intensive training; 2) Propose and implement Including man-machine fusion collaboration, field large-scale real-time communication and embedded computing of agricultural machinery dynamic formation, collaborative work mode;
(3) Research team and work foundation
The lead unit of Project 3, Tsinghua University, relies on the Institute of Navigation and Control, Department of Automation, Tsinghua University. It is mainly driven by the core technology of intelligent unmanned systems, and has undertaken and completed a number of scientific research projects in the fields of unmanned aerial vehicles, unmanned vehicles, space robots, and underwater unmanned systems. The participating unit Heilongjiang Wo Technology Development Co., Ltd. is a national high-tech enterprise specializing in creating high-end agricultural machinery installation and agricultural services in China. Has a number of patented technologies and won a number of scientific and technological awards.
Here are some on-site scientific research projects of participating units, as well as related agricultural machinery and tools.
In terms of previous technology accumulation, a high-dynamic UAV group communication networking system has been realized, which can meet the requirements of group agricultural machinery on-site communication; the distributed cooperative control and optimization mechanism in biological populations has been proposed and verified, and a variety of unmanned and Algorithms related to mission planning and cooperative control.
During the task application and start-up process, Liu Sheng's investigation on the standardized operation of agricultural machinery in paddy and dry fields was carried out together with other subjects of the project.
On-site inspection of the project results application demonstration base: 290 farm on-site network, electricity and farmland facilities.
(4) Project schedule and assessment methods
This table gives a three-year, quarterly schedule. The third topic is carried out in parallel according to three main lines; the first one is about the simulation and verification platform design of the cooperative operation algorithm of agricultural machinery; the second one is the verification design and finalization of the computing hardware platform for realizing swarm intelligent control; Reports, papers, patents, industry norms, and personnel training generated by the technical summary. Finally, the system integration and application demonstration will be completed with the overall project.
The content of each part of the assessment method of the project is concentrated in this table.
(5) Project budget and expected results
This table gives the total funding and budget allocation for the project.
The expected results include three aspects: the first aspect is the algorithm software and hardware platform of swarm intelligence; the second part is the realization of system integration and demonstration, including the autonomous decision-making and task assignment of more than three types of unmanned agricultural machines; the realization of more than 5 sets The third part is the content of related papers, patents and soft works. Finally, three industry standards of agricultural intelligence, autonomy and unmannedization will be formed.
The above is the content of my report, please criticize and correct me.
