Search

CN-121995916-A - Navigation and obstacle avoidance method and system between irrigation robots

CN121995916ACN 121995916 ACN121995916 ACN 121995916ACN-121995916-A

Abstract

The invention provides a method and a system for navigation and obstacle avoidance between irrigation robots, and belongs to the technical field of high-precision navigation. The invention aims to solve the problems of path planning and execution caused by dynamic barriers, sudden animal interference and environmental changes. The method comprises the steps of obtaining the current robot pose, judging the motion trail of a dynamic obstacle, generating an execution path of an irrigation robot, generating a standby navigation sequence, generating the execution path of the irrigation robot, generating a prediction result of the motion trail of a sudden animal, and generating a safe irrigation execution instruction. The system comprises a path planning module, a low-power consumption processor, a navigation instruction generating module, a path adjusting module, a safe distance verification module, a track smoothing module, a real-time feedback module, a feature extraction module, an image processing accelerator, a track prediction analysis module, a path analysis module, an obstacle avoidance decision module, an irrigation sequence optimizing module, a distance adjusting module and a track verification module.

Inventors

  • LIU YONG
  • Wu Qinsong
  • XU WENCHAO
  • ZHANG MEILI
  • LI XINYI
  • CHEN TING
  • YU HAIYANG
  • WU HAIFENG

Assignees

  • 哈尔滨东水智慧农业科技开发有限公司

Dates

Publication Date
20260508
Application Date
20260129

Claims (10)

  1. 1. An irrigation robot inter-robot navigation and obstacle avoidance method, comprising: Step 1, collecting field real-time data, generating a preliminary environment map, integrating the preliminary environment map by adopting a sensor fusion engine, determining the current pose of an irrigation robot and acquiring the real-time environment map; step 2, extracting local area visual data according to the current pose of the irrigation robot, preprocessing the local area visual distance data, and detecting the preprocessed local area visual data through a pose computing component to obtain a motion track of a dynamic obstacle; Step 3, if the motion trail of the dynamic obstacle is overlapped with the irrigation robot route, generating an alternative obstacle avoidance route through a route generation module, adjusting the navigation instruction, correcting the adjusted navigation instruction based on the environment data in the real-time environment map, and generating an execution route of the irrigation robot; Step 4, if the execution path of the irrigation robot does not meet the low-delay response, updating the path through a rapid iterative optimization algorithm to obtain a safe irrigation execution sequence, predicting a dynamic obstacle track in the safe irrigation execution sequence and calculating collision probability, and if the collision probability exceeds a threshold value, performing path adjustment to generate a standby navigation sequence; step 5, integrating the standby navigation sequence and the real-time environment map data and inputting the integrated navigation sequence and the real-time environment map data into a low-power processor to obtain an obstacle classification result, and performing safe distance maintenance verification on the obstacle classification recognition result by adopting a navigation instruction generation module to obtain a corrected execution path of the irrigation robot; Step 6, extracting real-time environment feature vectors of the execution paths of the corrected irrigation robots, compensating muddy ground interference in the real-time environment feature vectors to obtain compensated visual feature data, and carrying out dynamic track prediction on the compensated visual feature data through a pose calculation assembly to obtain a prediction result of the motion track of the sudden animal; Step 7, if the predicted movement track of the burst animal overlaps with the execution path of the corrected irrigation robot, generating an alternative obstacle avoidance path set aiming at the field rows based on the obstacle avoidance decision logic and determining a final field row navigation sequence; and 8, converting the final field inter-row navigation sequence into a navigation instruction which can be executed by the equipment to obtain a safe irrigation execution instruction.
  2. 2. The method of inter-irrigation robot navigation and obstacle avoidance as recited in claim 1 wherein the pose computing assembly comprises: a low power consumption processor, an image processing accelerator, a sensor combination and an inertial measurement unit; the sensor is used for collecting real-time data of the field environment; The image processing accelerator is used for compensating muddy ground interference in the environment characteristic vector; The inertial measurement unit is used for collecting inertial measurement unit data of the field environment; The pose computing component is used for extracting visual characteristic data and detecting the extracted visual characteristic data.
  3. 3. The method for inter-irrigation robot navigation and obstacle avoidance as recited in claim 1, wherein step 1 specifically comprises: acquiring real-time data from a field environment through a sensor combination to obtain original environment data, wherein the original environment data comprises laser radar point clouds, camera images and inertial measurement unit data; detecting a missing value of the preliminary environmental map, and if the data in the preliminary environmental map is missing, supplementing a missing area through an interpolation method to obtain a complete preliminary environmental map; integrating the completed preliminary environment map according to the sensor fusion engine, and fusing inertial measurement unit data to compensate GNSS signals to obtain an optimized environment map; And if the current pose deviation of the irrigation robot exceeds a first preset threshold, optimizing the pose by adopting a particle filtering algorithm to obtain a high-precision pose, updating the optimized environment map according to the high-precision pose, and obtaining a real-time environment map.
  4. 4. The method for inter-irrigation robot navigation and obstacle avoidance as recited in claim 1, wherein step 2 comprises: Extracting local area visual data from real-time data acquired by a sensor based on the current pose of the irrigation robot, denoising the local area visual data to obtain first visual data, and correcting illumination change of the first visual data by adopting an image processing accelerator to obtain second visual data; The second visual data is subjected to feature extraction through the pose computing component, the crop row structure in the field is detected, the crop distribution situation is obtained, and if a non-crop row area exists in the crop distribution situation, the position of the hard obstacle is judged through an edge detection algorithm; And analyzing the second visual data by adopting an optical flow algorithm based on the hard obstacle position to obtain the motion trail of the dynamic obstacle.
  5. 5. The method for inter-irrigation robot navigation and obstacle avoidance as recited in claim 1, wherein step 3 comprises: 3.1, if the movement speed of the dynamic obstacle exceeds a second preset threshold, predicting the future position of the dynamic obstacle by adopting a Kalman filtering algorithm, and updating the set navigation path of the irrigation robot according to the predicted position of the obstacle to generate a dynamic navigation instruction; Step 3.2, if the motion track of the dynamic obstacle is overlapped with the navigation path of the updated irrigation robot, calculating an alternative obstacle avoidance path through a path planning module to obtain an alternative path set, and if the motion track of the dynamic obstacle is not overlapped with the navigation path of the updated irrigation robot, directly converting the navigation path of the updated irrigation robot into a navigation instruction executable by equipment to obtain a safe irrigation execution instruction; step 3.3, adopting real-time data acquired by a sensor combination to carry out environment constraint analysis on the alternative path set to obtain a screened path set; step 3.4, integrating the screened path set and the motion trail of the dynamic obstacle through a fusion algorithm to obtain an optimal obstacle avoidance path, if the optimal obstacle avoidance path meets the preset passing condition, adjusting the navigation path of the irrigation robot according to the optimal obstacle avoidance path, and if the optimal obstacle avoidance path does not meet the preset passing condition, repeating the steps 3.1-3.3 until the preset passing condition is met; and 3.5, correcting the adjusted navigation path of the irrigation robot in real time by adopting original environment data to obtain corrected navigation instructions, updating motion control parameters of the irrigation robot by the corrected navigation instructions, determining a final field inter-row navigation instruction, and generating a driving signal according to the final inter-row navigation instruction to obtain an execution path of the irrigation robot.
  6. 6. The method for inter-irrigation robot navigation and obstacle avoidance as recited in claim 1, wherein step 4 comprises: If the execution path of the irrigation robot coincides with the motion track of the dynamic obstacle, calculating an alternative path set through a path planning module to obtain a preliminary path set; Performing environment constraint analysis on the preliminary path set by adopting real-time sensor data, screening out an optimized path set meeting traffic conditions, evaluating potential interference in the optimized path set by using a dynamic track prediction algorithm, and determining an optimal obstacle avoidance path; Fusing the optimal obstacle avoidance path with the acquired real-time data to generate an adjusted irrigation sequence, and performing low-delay response verification on the adjusted irrigation sequence by adopting a rapid iteration optimization algorithm to obtain a safe execution sequence; And generating a driving signal according to the safety execution sequence, determining a final irrigation control instruction, and dynamically correcting the final irrigation control instruction through real-time data acquired by a sensor combination to obtain a standby navigation sequence.
  7. 7. The method for inter-irrigation robot navigation and obstacle avoidance as recited in claim 1, wherein step 5 comprises: integrating the standby navigation sequence and the real-time environment map data, and inputting the integrated navigation sequence and the real-time environment map data into a low-power processor for feature extraction and classification processing to obtain an obstacle classification result; Inputting the obstacle classification result into a navigation instruction generation module to obtain an initial navigation instruction sequence, and if the initial navigation instruction sequence is overlapped with the obstacle position in the obstacle classification result, recalculating an alternative path through a path adjustment module to obtain an adjusted navigation instruction sequence; Performing distance constraint analysis on the adjusted navigation instruction sequence through a safety distance verification module to obtain a navigation instruction sequence meeting the safety distance, and performing optimization processing on the navigation instruction sequence meeting the safety distance through a track smoothing module to obtain a smooth inter-row path planning; and dynamically correcting the smooth inter-row path planning through a real-time feedback module to obtain the corrected execution path of the irrigation robot.
  8. 8. The method for inter-irrigation robot navigation and obstacle avoidance as recited in claim 1, wherein step 6 comprises: Extracting real-time environment feature vectors of the corrected execution paths of the irrigation robots through a feature extraction module, and compensating muddy ground interference in the environment feature vectors through an image processing accelerator to obtain compensated visual feature data; carrying out dynamic pose estimation on the compensated visual characteristic data by adopting a pose calculation component to obtain dynamic pose parameters, and carrying out motion trail prediction on the dynamic pose parameters by using a trail prediction analysis module to obtain predicted motion trail data; And acquiring a corresponding sensor acquisition image according to the predicted motion trail data, detecting an animal target in the image by adopting a YOLOv s model, extracting an animal target boundary box, tracking the motion of the animal by Kalman filtering, and predicting the motion trail of the burst animal.
  9. 9. The method for inter-irrigation robot navigation and obstacle avoidance as recited in claim 1, wherein step 7 comprises: If the predicted movement track of the burst animal is overlapped with the execution path of the corrected irrigation robot, extracting track features of the overlapped region through a path analysis module to obtain feature data of the overlapped region; processing the feature data of the overlapping area through an obstacle avoidance decision module to generate alternative obstacle avoidance paths among the field rows, and obtaining an alternative path set; Integrating the alternative path set with the environment map data by adopting a path fusion algorithm to obtain fused path data, and performing sequence adjustment on the fused path data by using an irrigation sequence optimization module to obtain an optimized irrigation execution sequence; if the optimized irrigation execution sequence and the preset barrier safety distance do not meet the preset distance, calculating path offset through a distance adjustment module to obtain adjusted safety path data; And carrying out feasibility analysis on the adjusted safety path data by adopting a track verification module to obtain a final field inter-row navigation sequence.
  10. 10. An inter-irrigation robot navigation and obstacle avoidance system applied to an inter-irrigation robot navigation and obstacle avoidance method as claimed in any one of claims 1 to 9, comprising: The path planning module is used for planning an execution path of the irrigation robot and inputting an alternative obstacle avoidance path and a preliminary path; the navigation instruction generation module is used for generating a navigation instruction of the irrigation robot; the path adjusting module is used for adjusting the execution path of the irrigation robot; the safety distance verification module is used for carrying out distance constraint analysis on the navigation instruction of the irrigation robot; The track smoothing module is used for optimizing the navigation instruction sequences meeting the safety distance; The real-time feedback module is used for dynamically correcting the smooth inter-row path planning; the feature extraction module is used for extracting real-time environment feature vectors of the execution path of the irrigation robot; the track prediction analysis module is used for predicting the motion track of the irrigation robot according to the dynamic position parameters of the irrigation robot; the path analysis module is used for extracting track characteristics of an overlapping area of an execution path of the irrigation robot and a motion track of the burst animal; The obstacle avoidance decision module is used for analyzing the track characteristics of the overlapped area and generating alternative obstacle avoidance paths among the field rows; the irrigation sequence optimizing module is used for carrying out sequence adjustment on the path data; The distance adjustment module is used for adjusting the path offset of the irrigation execution sequence; the track verification module is used for carrying out feasibility analysis on the safety path data; the image processing accelerator is used for carrying out compensation processing on muddy ground interference in the environment characteristic vector; and the low-power-consumption processor is used for carrying out feature extraction and classification processing on the integrated data to output obstacle classification results.

Description

Navigation and obstacle avoidance method and system between irrigation robots Technical Field The invention relates to a method and a system for navigation and obstacle avoidance between irrigation robots, and belongs to the technical field of high-precision navigation. Background In modern agriculture, intelligent irrigation robots are considered as an important technical means for improving farmland production efficiency and resource utilization rate. Particularly in complex unstructured environments, such as scenes with complex terrain, various obstacles and unstable signals in farmlands, the realization of accurate navigation and safe obstacle avoidance becomes a key for promoting agricultural automation. The robot realizes high-efficiency irrigation, so that not only can the water resource distribution be optimized, but also the cost and error of manual operation can be reduced. However, current navigation and obstacle avoidance techniques face significant technical challenges in such dynamic and non-standardized environments in farmlands, limiting the wide application of irrigation robots in practical production. The existing methods have obvious limitations in coping with complex farmland environments. The traditional navigation technology depends on Global Navigation Satellite System (GNSS) signals, but the signals are unstable in farmland due to terrain shielding or vegetation interference, so that the high-precision positioning requirement is difficult to meet. Meanwhile, the obstacle avoidance method based on two-dimensional vision has poor effect in distinguishing crops from hard obstacles, and particularly under the condition of large illumination variation, the vision system easily misjudges the crops as the obstacles or recognizes the hard objects to be avoided as the crushable objects. The methods react slowly to rapidly changing obstacles in a dynamic environment, cannot meet the real-time requirement, and cause that the robot is difficult to work in a complex farmland with high efficiency. The core technical difficulty is focused on how to realize low-delay accurate navigation and safety obstacle avoidance by only relying on an onboard computing unit. First, the dynamics of the farm environment requires that the robot be able to complete the environment awareness and decision in less than 100 milliseconds. The limited computational power of the on-board computing unit makes real-time processing of complex algorithms difficult, especially when navigation and obstacle avoidance tasks need to be processed simultaneously, allocation of computing resources becomes a bottleneck. This low delay requirement further amplifies the second technical difficulty of how to accurately distinguish crops from hard obstacles in scenes where light changes and dynamic obstacles are present. For example, in farmlands, irrigation robots may need to identify whether the front is a low crop that can be crushed or a stone or person that needs to be evaded while moving quickly. Because the illumination changes can interfere with the data of the vision sensor, the robot can hit obstacles or erroneously bypass crops due to misjudgment, and the irrigation efficiency is affected. Therefore, how to realize accurate navigation and safe obstacle avoidance between rows with delay lower than 100 milliseconds only by means of an airborne computing unit in a complex farmland environment with unstable GNSS signals and frequent occurrence of dynamic obstacles becomes a key problem of technological breakthrough of intelligent irrigation robots. Disclosure of Invention The invention provides a navigation and obstacle avoidance method and system between irrigation robots, which aims to solve the problems of path planning and execution caused by dynamic obstacles, sudden animal interference and environmental changes. The technical scheme adopted for solving the problems is that the inter-irrigation robot navigation and obstacle avoidance method provided by the invention comprises the following steps: Step 1, collecting field real-time data, generating a preliminary environment map, integrating the preliminary environment map by adopting a sensor fusion engine, determining the current pose of an irrigation robot and acquiring the real-time environment map; step 2, extracting local area visual data according to the current pose of the irrigation robot, preprocessing the local area visual distance data, and detecting the preprocessed local area visual data through a pose computing component to obtain a motion track of a dynamic obstacle; Step 3, if the motion trail of the dynamic obstacle is overlapped with the irrigation robot route, generating an alternative obstacle avoidance route through a route generation module, adjusting the navigation instruction, correcting the adjusted navigation instruction based on the environment data in the real-time environment map, and generating an execution route of the irrigation robot; Step 4, if t