CN-122015877-A - Robot operation track planning method, equipment and medium

CN122015877ACN 122015877 ACN122015877 ACN 122015877ACN-122015877-A

Abstract

The invention discloses a robot operation track planning method, equipment and medium, relating to the technical field of operation track planning, comprising the steps of collecting visual positioning map building data, carrying out coordinate alignment and passable domain generation, and outputting a passable map; the method comprises the steps of conducting obstacle semantic classification on a passable map, fusing a dynamic target prediction construction time occupation zone and a time feasible domain, conducting global coverage sequence planning on a compensation contract binding diagram to generate a global track fragment sequence, conducting queue insertion planning according to the compensation contract binding diagram according to a to-be-compensated queue to generate a compensation track, driving a robot to conduct compensation and acceptance, and outputting an operation report. The method generates the passable map through visual positioning map construction and coordinate alignment, and then establishes the time occupation zone and the time feasibility domain through obstacle semantic classification fusion dynamic target prediction, so as to form the process control capability of through planning, execution and acceptance, and improve the coverage integrity, continuous operation and delivery reliability.

Inventors

WU JIAN
MA YE
WANG FEI
YUAN MENGRU
SHI MIAO

Assignees

杭州它人机器人技术有限公司

Dates

Publication Date: 20260512
Application Date: 20260414

Claims (10)

1. A robot job trajectory planning method, comprising: collecting visual positioning map building data, carrying out coordinate alignment and passable domain generation, and outputting a passable map; performing obstacle semantic classification on the passable map, and fusing a dynamic target prediction construction time occupation zone and a time feasible domain; Performing overlay partition decomposition on the time feasible domain to generate an overlay partition set, simultaneously establishing a compensation contract binding diagram for the overlay partition set, and initializing an overlay quality field target layer; carrying out global coverage sequence planning on the compensation contract binding diagram to generate a global track fragment sequence; track tracking and local diversion are executed based on the global track segment sequence, and the overlay quality field is updated online to trigger skipping segments and write into a queue to be complemented; And performing queue insertion planning according to the to-be-compensated queue and the binding diagram of the compensation contract, generating a compensation track, driving a robot to perform compensation and acceptance, and outputting an operation report.
2. The robot job trajectory planning method according to claim 1, wherein the steps of collecting visual positioning map data, performing coordinate alignment and passable domain generation, and outputting a passable map are as follows: The robot calls a camera and inertial measurement to acquire continuous images and inertial data, synchronously acquires odometer information, forms visual positioning map building data, performs time synchronization and abnormal frame rejection on the visual positioning map building data, and extracts positioning characteristic information to obtain a continuous pose estimation result; registering the continuous pose estimation result with a working coordinate system to form an alignment environment map, identifying an obstacle region of the alignment environment map, performing expansion processing on the obstacle region to generate a passable domain, fusing the passable domain with a working boundary, and outputting a passable map.
3. The robot job trajectory planning method according to claim 2, wherein the obstacle semantic classification is performed on the passable map, and the time occupied zone and the time feasible domain are constructed by fusion of dynamic target prediction, specifically comprising the following steps: Reading a passable map, extracting a barrier region boundary, forming a candidate barrier set, performing geometric feature calculation and barrier semantic classification on the candidate barrier set, and generating a barrier semantic tag; Generating a welt buffer area for the wall body obstacle according to the obstacle semantic tag, generating a detour buffer area for the columnar obstacle, forming semantic safety constraint, collecting current position and speed information of the robot, and recursively predicting the position according to time steps in a prediction time domain to generate a prediction occupied area; and mapping the predicted occupied area and the semantic security constraint to the passable map together to construct a time occupied zone, performing time domain clipping on the passable map based on the time occupied zone, and outputting a time feasible domain.
4. The method for planning a track of a robot operation according to claim 1, wherein the step of performing coverage partition decomposition on the time feasible region to generate a coverage partition set, and simultaneously establishing a reconvergence contract binding diagram for the coverage partition set comprises the following specific steps: Extracting regional boundary information of a time feasible region to form a region to be decomposed, carrying out connectivity analysis on the region to be decomposed, executing covering partition decomposition to form a plurality of covering partitions, and summarizing to form a covering partition set; Extracting an accessible boundary and an withdrawable boundary from the coverage partition set, calculating a coverage main direction and a steering buffer range to obtain coverage partition planning constraint, and establishing a coverage partition adjacency relationship based on a shared boundary between the coverage partition sets to form a coverage partition switching diagram; And selecting a compensation entry point and a compensation exit point on the accessible boundary and the withdrawable boundary of the overlay partition based on the overlay partition planning constraint, setting a compensation triggering condition and a compensation ending condition, generating a compensation contract, and storing the compensation contract and the overlay partition switching diagram in a correlated manner to form a compensation contract binding diagram.
5. The robot job trajectory planning method of claim 1, wherein the coverage quality field target layer is obtained by establishing a target quality distribution layer consistent with a space range of a time-feasible domain under a passable map coordinate system, writing the target quality distribution layer by position by a target coverage strength and a target overlapping level of a corresponding job type in the job quality requirements to form a basic target value, mapping a welt buffer area corresponding to a wall obstacle and a bypass buffer area corresponding to a columnar obstacle to the target quality distribution layer by combining an obstacle semantic tag, and writing the corresponding target quality requirements in the welt buffer area and the bypass buffer area.
6. The method for planning a track of a robot operation according to claim 4, wherein the global coverage sequence planning is performed on the supplementary contract binding diagram to generate a global track segment sequence, and the specific steps are as follows: Calculating switching cost between the coverage partitions based on a coverage partition switching diagram in the reconversion contract binding diagram, fusing the switching cost with predicted coverage cost in the coverage partitions to form global planning cost, and generating a coverage partition access sequence by taking the coverage partition with the starting position as a starting point and through global planning cost constraint; and combining the overlay main directions of all overlay partitions in the overlay partition access sequence to generate track fragments in the partitions, generating switching track fragments based on the back-patch exit points and the back-patch entry points, connecting the track fragments in the partitions with the switching track fragments in series and associating corresponding back-patch contracts, and generating a global track fragment sequence.
7. The method for planning a track of a robot job according to claim 6, wherein the steps of performing track tracking and local diversion based on the global track segment sequence, and updating the overlay quality field on line to trigger skipping segments and writing the skip segments into the queue to be complemented are as follows: Reading a global track segment sequence, sequentially scheduling current track segments to be executed, generating a motion control instruction based on the current pose state of the robot and the current track segments to be executed, and tracking the track; continuously acquiring environment observation data in the track tracking process, and performing consistency check with a time occupation zone to generate a local rerouting track; And continuously acquiring operation feedback data in the track execution process, mapping the operation feedback data to an operation area, obtaining a coverage quality field quality value, comparing and updating the coverage quality field through a coverage quality field target layer and the coverage quality field quality value, and writing the track segment to be executed currently into a queue to be complemented.
8. The method for planning the operation track of the robot according to claim 7, wherein the step of performing the queue insertion planning according to the binding diagram of the compensation contract according to the queue to be compensated, generating the compensation track, driving the robot to perform compensation and acceptance, and outputting the operation report comprises the following specific steps: Reading a queue to be complemented, acquiring a complemented entry point and a complemented exit point associated with a skipped fragment, and simultaneously reading a complemented binding diagram to acquire a covered partition set and a covered partition switching diagram associated with the skipped fragment; Calculating the queue inserting cost from the current position of the robot to the re-compensation entry point based on the re-compensation contract binding diagram, generating a queue inserting planning result, and selecting a skip segment corresponding to the queue inserting planning result as a current re-compensation task; Generating a compensation track based on a compensation entry point and a compensation exit point of the current compensation task, inserting a global track segment sequence to form an insertion queue execution sequence, driving a robot to execute the compensation track, generating a check conclusion based on a coverage quality field and a coverage quality field target layer, and outputting a job report.
9. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the robot job trajectory planning method according to any one of claims 1-8.
10. A computer readable storage medium having stored thereon a computer program, characterized in that the computer program when executed by a processor realizes the steps of the robot job trajectory planning method according to any one of claims 1 to 8.

Description

Robot operation track planning method, equipment and medium Technical Field The invention relates to the technical field of operation track planning, in particular to a robot operation track planning method, equipment and medium. Background The track planning technology based on Visual synchronous positioning and mapping (Visual SLAM) has become a core means for realizing environment perception and path generation, the existing mainstream technical scheme generally relies on depth cameras or laser radars to acquire point cloud data, a three-dimensional map of a static environment is constructed through feature matching and pose estimation, a passable area is divided by using a rasterization method on the basis, and then a planning algorithm mostly adopts a full Coverage Path Planning (CPP) strategy, such as Niu Geng decomposition or spanning tree coverage algorithm, to divide a working space into a plurality of subareas, and a global path traversing all subareas is generated according to heuristic rules. As application scenarios extend to highly dynamic, unstructured and complex semantic environments, the prior art gradually reveals limitations in terms of coverage integrity under processing space-time coupling constraints, especially there is an optimization space in terms of process control logic between high-level decisions and underlying execution, traditional planning methods often treat environmental modeling as a static process, or simply process dynamic obstacles as instantaneous forbidden regions, lack a feasible domain construction mechanism in terms of predictive modeling and temporal dimensions of dynamic target motion trends, and when moving objects or temporary obstacles exist in the environment, the process control strategy of the existing scheme has difficulty in prejudging its influence on coverage quality in the planning phase, resulting in frequent encountering local path blockage during execution of robots. Disclosure of Invention The present invention has been made in view of the above-described problems occurring in the prior art. Therefore, the invention provides a robot operation track planning method which solves the problem that coverage integrity is difficult to predict in a planning stage and control stably in an execution stage. In order to solve the technical problems, the invention provides the following technical scheme: The invention provides a robot operation track planning method, which comprises the steps of collecting visual positioning map building data, carrying out coordinate alignment and passable domain generation, outputting a passable map, carrying out obstacle semantic classification on the passable map, fusing dynamic target prediction construction time occupation zones and time feasible domains, carrying out coverage partition decomposition on the time feasible domains, generating a coverage partition set, simultaneously establishing a compensation contract binding map for the coverage partition set, initializing a coverage quality field target layer, carrying out global coverage sequence planning on the compensation contract binding map, generating a global track fragment sequence, carrying out track tracking and local diversion on the basis of the global track fragment sequence, carrying out on-line update of coverage quality field trigger skip fragments, writing into a to-be-compensated queue, carrying out insertion planning according to the compensation contract binding map according to the to-be-compensated queue, generating a compensation track, driving a robot to carry out compensation and inspection, and outputting an operation report. The invention relates to a robot operation track planning method, which comprises the following steps of collecting visual positioning map building data, carrying out coordinate alignment and passable domain generation, and outputting a passable map: The robot calls a camera and inertial measurement to acquire continuous images and inertial data, synchronously acquires odometer information, forms visual positioning map building data, performs time synchronization and abnormal frame rejection on the visual positioning map building data, and extracts positioning characteristic information to obtain a continuous pose estimation result; registering the continuous pose estimation result with a working coordinate system to form an alignment environment map, identifying an obstacle region of the alignment environment map, performing expansion processing on the obstacle region to generate a passable domain, fusing the passable domain with a working boundary, and outputting a passable map. The invention relates to a robot operation track planning method, which is a preferable scheme, wherein the obstacle semantic classification is carried out on a passable map, a time occupation zone and a time feasible domain are built by fusion of dynamic target prediction, and the method comprises the following specific steps: Reading a passabl