CN-122018519-A - Unmanned aerial vehicle interactive complex path planning method based on large model

CN122018519ACN 122018519 ACN122018519 ACN 122018519ACN-122018519-A

Abstract

The invention discloses an unmanned aerial vehicle interactive complex path planning method based on a large model, which comprises the following steps of selecting a conversational large model, calling a cloud voice recognition service to obtain a user voice instruction text, establishing an unmanned aerial vehicle knowledge base for communication, defining the large model as an operation assistant of an unmanned aerial vehicle and a rule for complex path planning, carrying out semantic analysis and reasoning on a natural language instruction through the large model based on a preset unmanned aerial vehicle knowledge base to generate a structured flight control instruction sequence, and using a multithreading architecture for establishing and maintaining communication connection with unmanned aerial vehicle flight control and processing interaction between a user and the large model. The invention realizes the interactive planning of the complex path of the unmanned aerial vehicle and promotes the popularization of the technical development and the use of the unmanned aerial vehicle.

Inventors

SONG YANGUO
CHEN YAN
WANG HUANJIN

Assignees

南京航空航天大学

Dates

Publication Date: 20260512
Application Date: 20251212

Claims (6)

1. The unmanned aerial vehicle interactive complex path planning method based on the large model is characterized by comprising the following steps of: S1, selecting a dialogue type large model, and accessing a cloud voice recognition service; s2, establishing an unmanned aerial vehicle knowledge base for communication; S3, configuring the large model as an operation assistant of the unmanned aerial vehicle, and setting rules of complex path planning; S4, carrying out semantic analysis and reasoning on natural language instructions through the large model based on a preset unmanned aerial vehicle knowledge base to generate a structured flight control instruction sequence; S5, establishing and maintaining communication connection with unmanned aerial vehicle flight control based on a multithreading architecture, and processing interaction between a user and the large model.
2. The unmanned aerial vehicle interactive complex path planning method according to claim 1, wherein the speech recognition service in step S1 is an online recognition language service.
3. The unmanned aerial vehicle interactive complex path planning method according to claim 1, wherein the unmanned aerial vehicle knowledge base in step S2 comprises: 1) Aiming at the condition of undefined instructions, rules for actively inquiring flight distance, angle and altitude parameters of a user when the large model controls the unmanned aerial vehicle are specified; 2) Rules that specify that large models only allow for invoking predefined drone control functions; 3) The function definition for the normal operation of the unmanned aerial vehicle at least comprises an unlocking function, a take-off function, a positioning function for acquiring the current position, a waypoint flight function, a hovering waiting function, a landing function, a yaw function setting, a yaw angle acquisition function, a return function, a flight function according to a preset track, a back-and-forth and left-and-right movement function, a flight mode switching function and a flight speed adjustment function; 4) Dialog templates and examples of complex path planning; 5) The execution logic of the complex path planning comprises an instruction generation and issuing sequence after the path planning is completed; 6) Defining a height coordinate, and representing the position of the unmanned aerial vehicle by adopting a North east coordinate system; 7) Unmanned aerial vehicle control function collection based on unmanned aerial vehicle communication protocol is realized to encapsulate it into a class, supply the big model to carry out the function call.
4. The unmanned aerial vehicle interactive complex path planning method based on the large model of claim 1, wherein in step S3, the large model is configured as an unmanned aerial vehicle operation assistant in a mode of presetting prompt information, and when a user task instruction is received, all reasoning behaviors of the large model are based on the unmanned aerial vehicle knowledge base.
5. The unmanned aerial vehicle interactive complex path planning method based on the large model of claim 1, wherein in step S4, the large language model calls a function preset in the unmanned aerial vehicle knowledge base to generate a structured flight control instruction sequence according to a natural language instruction input by a user, and interprets the flight control instruction sequence in natural language for the user to confirm.
6. The unmanned aerial vehicle interactive complex path planning method according to claim 1, wherein in step S5, the multithreading architecture comprises: The system comprises a first thread, a second thread, a third thread and a fourth thread, wherein the first thread designates a fixed port during initialization, establishes TCP communication connection with the unmanned aerial vehicle flight control, updates the unmanned aerial vehicle state in a circulation at fixed time and processes communication with the unmanned aerial vehicle, and receives and recognizes voice input of a user in the circulation, sends a text obtained through recognition to the large model and receives response of the large model.

Description

Unmanned aerial vehicle interactive complex path planning method based on large model Technical Field The invention provides an unmanned aerial vehicle interactive complex path planning method based on a large model, and belongs to the technical field of unmanned aerial vehicles. Background An unmanned aerial vehicle, or unmanned aerial vehicle, is a unmanned aerial vehicle controlled by a radio remote control device or a built-in program. In recent years, along with the development of sensor technology, flight control algorithm and communication technology, unmanned aerial vehicles are widely applied to the fields of military, agriculture, logistics, disaster relief, environmental monitoring and the like. Traditional unmanned aerial vehicle mainly relies on manual remote control operation, and operating personnel passes through ground control station real-time control unmanned aerial vehicle's speed and plans unmanned aerial vehicle's flight path, and this kind of mode is higher to operating personnel's skill requirement. With the rapid development of artificial intelligence and machine learning technologies, unmanned aerial vehicles gradually develop from simple remote control operations to intelligent and autonomous directions. Modern unmanned aerial vehicles are able to collect environmental data via sensors, cameras and other devices and utilize algorithms for real-time analysis and decision making. In order to reduce the use threshold and difficulty of an unmanned aerial vehicle and promote the development of unmanned aerial vehicle technology, the invention provides an unmanned aerial vehicle interactive complex path planning method based on a large model. Disclosure of Invention In order to solve the technical problems, the invention aims to provide the unmanned aerial vehicle interactive complex path planning method based on the large model, aiming at complex path planning, the unmanned aerial vehicle interactive complex path planning method cannot be directly drawn through single waypoint coordinates or simple straight line and arc commands, so as to promote the development of unmanned aerial vehicle technology. The invention provides a large model-based unmanned aerial vehicle interactive complex path planning method, which comprises the following steps: S1, selecting a dialogue type large model, and accessing a cloud voice recognition service; s2, establishing an unmanned aerial vehicle knowledge base for communication; S3, configuring the large model as an operation assistant of the unmanned aerial vehicle, and setting rules of complex path planning; S4, carrying out semantic analysis and reasoning on natural language instructions through the large model based on a preset unmanned aerial vehicle knowledge base to generate a structured flight control instruction sequence; S5, establishing and maintaining communication connection with unmanned aerial vehicle flight control based on a multithreading architecture, and processing interaction between a user and the large model. Further, the voice recognition service in step S1 is an online recognition language service. Further, the unmanned aerial vehicle knowledge base in step S2 includes: 1) Aiming at the condition of undefined instructions, rules for actively inquiring flight distance, angle and altitude parameters of a user when the large model controls the unmanned aerial vehicle are specified; 2) Rules that specify that large models only allow for invoking predefined drone control functions; 3) The function definition for the normal operation of the unmanned aerial vehicle at least comprises an unlocking function, a take-off function, a positioning function for acquiring the current position, a waypoint flight function, a hovering waiting function, a landing function, a yaw function setting, a yaw angle acquisition function, a return function, a flight function according to a preset track, a back-and-forth and left-and-right movement function, a flight mode switching function and a flight speed adjustment function; 4) Dialog templates and examples of complex path planning; 5) The execution logic of the complex path planning comprises an instruction generation and issuing sequence after the path planning is completed; 6) Defining a height coordinate, and representing the position of the unmanned aerial vehicle by adopting a North east coordinate system; 7) Unmanned aerial vehicle control function collection based on unmanned aerial vehicle communication protocol is realized to encapsulate it into a class, supply the big model to carry out the function call. Further, in step S3, the large model is configured as an unmanned aerial vehicle operation assistant in such a way that prompt information is preset, and when a user task instruction is received, all reasoning behaviors of the large model are based on the unmanned aerial vehicle knowledge base. Further, in step S4, the large language model invokes a function preset in the unmanned aerial vehi