CN-121657741-B - Automatic calculation and compensation method and system for dynamic striking delay of laser weeding robot

CN121657741BCN 121657741 BCN121657741 BCN 121657741BCN-121657741-B

Abstract

The invention discloses a method and a system for automatically calculating and compensating dynamic striking delay of a laser weeding robot, which belong to the technical field of agricultural intelligent equipment and are realized by correcting laser galvanometer, a camera and cooperative parameters of the laser galvanometer and the camera of a robot system in a static calibration stage, and eliminating nonlinear errors of laser scanning, imaging distortion errors of the camera and pose and mapping errors between a galvanometer coordinate system and a camera coordinate system; in a dynamic test stage, a robot advances at a constant speed, laser striking is performed on wood chips marked in advance, the position deviation between an actual striking point and a theoretical target point is obtained through an image recognition technology, the delay time of a system is reversely pushed based on the deviation and the speed of the robot, compensation parameters are iteratively updated, and finally the compensation parameters are solidified to a control system after the iterative optimization striking deviation is smaller than the set precision.

Inventors

ZHAO ZHIWU
WANG HONGKUN
GUO LIANG
XU JIANWEI
Zhai Jiacan

Assignees

蔚蓝引擎(上海)科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260206

Claims (5)

1. The automatic calculation and compensation method for the dynamic striking delay of the laser weeding robot is characterized by being implemented in the following way, S1, a static calibration stage; Calibrating inherent parameters of a robot system in a static state, and correcting nonlinear errors of a laser galvanometer, distortion errors of a camera and pose errors of galvanometer-camera installation to obtain a static precision reference; S2, a dynamic test stage; The robot moves at a constant speed v, identifies and tracks the wood chip with the marking point, starts a striking program, strikes the marking center of the wood chip by laser, detects the position of the wood chip, the corner point of the wood chip and the center of the black spot by an image identification module, calculates to obtain the pixel position P actual of the striking black spot and the pixel position P theory of the theoretical target center, and then calculates the space deviation delta d between the actual striking point and the theoretical target center; s3, an iterative optimization stage; When the striking deviation exceeds a set threshold, calculating a time delay compensation parameter, and calculating a delay delta t=delta d/v according to the travelling speed v of the robot, updating a system time compensation parameter, wherein an initial compensation parameter T old is set to 0, and then carrying out iterative updating according to the time delay delta T calculated by each dynamic striking, wherein the time delay compensation parameter T new =T old +delta T is updated; s4, a parameter curing stage; and writing the final compensation value into a control system parameter area for subsequent automatic validation, and restarting the dynamic calibration flow through a one-key calibration function.
2. The automatic calculation and compensation method for dynamic striking delay of laser weeding robot according to claim 1, wherein a square wood chip with a fixed size is used as a dynamic calibration target carrier, target marks with fixed positions, such as center positions, are preset on the wood chip, the program performs laser striking on the target marks after detecting the targets in the traveling process, the striking deviation is calculated under the local coordinate system of the wood chip by detecting the actual striking black spot position formed on the wood chip after laser striking, and the system time delay is reversely deduced by combining the robot advance speed, so that the self-learning and convergence of delay compensation parameters are realized.
3. The automatic calculation and compensation method for dynamic striking delay of laser weeding robot according to claim 1, wherein the calculation and compensation flow is as follows, Firstly, detecting four corner pixel points of wood chips; The four corner pixel coordinates are expressed as: ; secondly, detecting a black spot center pixel point; the black spot center pixel coordinates are expressed as: ; Thirdly, mapping the actual coordinates of the pixels to the wood chips; The length and width of the wood chips are L x and the width of the wood chips is L y ; Normalization coefficient: ; The black spot is relative to the central position of the wood chip, firstly defining the pixel coordinate (u c ,v c ) of the central position of the wood chip; ; calculating the position (X hit ,Y hit ) of the black spot relative to the center; ; step four, calculating the actual space deviation; the calculation formula of the actual spatial deviation Δd is: ; fifthly, obtaining time delta t required to be compensated; ; where v is the robot travel speed.
4. The automatic calculation and compensation method for dynamic striking delay of laser weeding robot according to claim 3, wherein Δd calculated based on the above method is the actual position deviation of the target point from the striking point, regardless of the direction angle of wood chip placement.
5. A system for automatically calculating and compensating for the dynamic striking delay of a laser weeding robot according to any one of claims 1 to 4, comprising the following cooperating modules: The visual perception and imaging module comprises an industrial camera and a light supplementing unit, wherein the industrial camera is a high-frame-rate and high-pixel industrial camera and is used for acquiring continuous image data of an operation area, the light supplementing unit adopts a high-brightness light source triggered synchronously with the camera and is used for inhibiting ambient light change and reflection interference and improving imaging stability of target wood chips and marking points thereof, and the visual perception and imaging module is used for identifying, tracking and beating the target wood chips in a dynamic scene and identifying and extracting sub-pixel level characteristic points for beating and positioning and black spots left after beating; The motion sensing module comprises an IMU inertial measurement unit and a wheel speed encoder, and is used for acquiring pose information of the laser weeding robot in a motion process in real time, outputting motion speed parameters of the robot under a world coordinate system and providing motion priori data for dynamic delay calculation; the laser vibrating mirror striking execution module comprises a laser, a double-shaft high-speed vibrating mirror and a laser control board card, wherein the double-shaft high-speed vibrating mirror is used for realizing high-speed deflection positioning of laser spots in a target plane, the laser is used for outputting a striking high-energy laser beam, and the laser control board card is integrated with a high-speed control unit and used for realizing accurate time sequence control of laser pulses and scanning control of the vibrating mirror and supporting microsecond-level response; The central computing and controlling module is an industrial personal computer or an embedded PC platform and is used for running a visual reasoning, motion prediction and delay compensation control program, wherein the central computing and controlling module deploys a target recognition and angular point and black spot detection algorithm to conduct real-time reasoning on image data collected by a camera, calculates total system delay from image collection, data processing to actual laser striking according to a detected striking error result and speed information output by a motion sensing module, and generates corresponding time compensation parameters based on the total delay.

Description

Automatic calculation and compensation method and system for dynamic striking delay of laser weeding robot Technical Field The invention relates to the field of agricultural intelligent equipment, in particular to a time delay automatic compensation method and a system for a laser weeding robot in a dynamic target beating process, which are used for improving the beating precision of the robot in a moving operation state. Background The existing laser weeding robot is often influenced by system delay when identifying targets (such as weeds) and striking the laser in the travelling process. Program execution laser emission instructions are based on time stamps and corresponding target space positions when the laser emission control instructions are executed, a time difference delta t exists between the laser emission control instructions and the actual execution of laser, so even if the laser is hit based on t time and the position (x, y, z) of t time, the robot still moves in the time period, when the laser is executed, the laser hit position is shifted relative to the target, delta d=delta t x v, the hit point is delayed from the target point theoretically, and therefore the factor of delta t needs to be taken into consideration in an execution instruction part, and prediction is advanced, so that the laser just hits the target position when the laser is executed. Such delays result from superposition of multiple links, including image acquisition delays, control signal transmission delays, laser execution delays, and vehicle motion offsets. At present, a system usually depends on static calibration, and can not accurately compensate for the delay in dynamic operation, so that the striking deviation is larger. Disclosure of Invention The invention overcomes the defects in the prior art, and provides a dynamic self-calibration delay compensation method and a dynamic self-calibration delay compensation system for a laser weeding robot. In order to solve the technical problems, the technical proposal adopted by the invention is that the automatic calculation and compensation method for the dynamic striking delay of the laser weeding robot is implemented according to the following mode, S1, a static calibration stage; And (3) calibrating inherent parameters of the robot system in a static state, wherein the inherent parameters comprise nonlinear error correction parameters of the laser galvanometer, distortion error correction parameters of the camera and pose error correction parameters of the galvanometer-camera installation, so that the theoretical target position under the galvanometer coordinate system can be accurately positioned through the target pixel position, and the accurate striking of laser is realized. S2, a dynamic test stage; The robot advances at a constant speed to dynamically strike the target wood chips entering the test area, the camera is used for detecting the corner points and the black spots of the set targets, the strike error is calculated, the time delay compensation parameters are calculated, and the system parameters are updated; s3, an iterative optimization stage; repeating the above process until the striking deviation of the continuous multiple tests is smaller than the set threshold value, and considering delay compensation convergence; s4, a parameter curing stage; and writing the final compensation value into a control system parameter area for subsequent automatic validation, and restarting the dynamic calibration flow through a one-key calibration function. Further, the static calibration method in S1 may be implemented by the following steps: s11, placing wood chips with center mark points; S12, controlling a laser to strike the mark points and recording the actual strike position and the theoretical position of the laser; s13, calculating nonlinear errors of the galvanometer, distortion errors of the camera and mounting pose errors of the galvanometer and the camera through recorded data; S14, updating system parameters, and realizing the correspondence between the target pixel position and the theoretical target position under the galvanometer coordinate system, thereby realizing the accurate striking of laser. Further, the dynamic test method in S2 is as follows: S21, the robot moves at a constant speed v; s22, identifying and tracking the wood chips with the marked points, and starting a striking program; S23, detecting the positions of the wood chips, the corner points of the wood chips and the centers of the black spots through an image recognition module, and calculating to obtain the pixel position P actual of actually striking the black spots and the pixel position P theory of the theoretical target center; S24, calculating the space deviation delta d (unit: cm) between the actual striking point and the theoretical target center; s25, calculating delay delta t=delta d/v according to the travelling speed v of the robot; S26, updating a system time compensation p