CN-122008250-A - Control method, system, device, medium, and program for humanoid robot

CN122008250ACN 122008250 ACN122008250 ACN 122008250ACN-122008250-A

Abstract

The invention relates to the technical field of robot control, and provides a control method, a system, equipment, a medium and a program of a humanoid robot, wherein the control method comprises the steps of controlling an environment sensing unit to scan a working area and generating first environment point cloud data representing a static map; the method comprises the steps of acquiring pose data of a target object by controlling a visual perception unit, controlling a force sense feedback unit integrated at the tail end of a mechanical arm executing mechanism, detecting the force sense feedback data of the mechanical arm executing mechanism in the process of contacting the target object, generating a motion track of the mechanical arm executing mechanism according to the pose data, first environment point cloud data and a preset operation strategy library, controlling the environment perception unit to scan an operation area in real time and generate second environment point cloud data representing dynamic obstacle distribution, and correcting the motion track in real time according to the second environment point cloud data to obtain an obstacle avoidance path. The invention is used for solving the problems of safe operation and high-precision operation of the humanoid robot in a dynamic change environment.

Inventors

KONG XIAOWEN
HE JUN
ZHANG YI
WANG CHENGYONG
LIN YINGYUN

Assignees

上海蜘蛛侠机器人有限公司

Dates

Publication Date: 20260512
Application Date: 20260410

Claims (13)

1. A control method of a humanoid robot, applied to a humanoid robot system including a visual sensing unit, an environment sensing unit, a mechanical arm executing mechanism and a force sense feedback unit, characterized in that the control method includes: controlling the environment sensing unit to scan a working area and generating first environment point cloud data representing a static map; The visual perception unit is controlled to acquire pose data of the target object; the force sense feedback unit is integrated at the tail end of the mechanical arm executing mechanism and used for detecting force sense feedback data of the mechanical arm executing mechanism in the process of contacting a target object; Judging whether to trigger an abnormal state according to the pose data and the force sense feedback data, identifying the arrangement type of a target material frame in the operation area according to the pose data when the abnormal state is not triggered, matching an unstacking strategy or a stacking strategy corresponding to the material frame arrangement type from a preset operation strategy library to serve as a current execution strategy, and selecting an abnormal processing strategy as the current execution strategy when the abnormal state is triggered; Based on the current execution strategy, calculating an end effector path and an articulation sequence of the mechanical arm by combining the space coordinates and the attitude angles of the target material frame in the pose data and the first environmental point cloud data, and generating a motion track of the mechanical arm executing mechanism; in the process of driving the mechanical arm executing mechanism to perform pose movement, controlling the environment sensing unit to scan a working area in real time and generating second environment point cloud data representing dynamic barrier distribution; the motion track is corrected in real time according to the second environmental point cloud data to obtain an obstacle avoidance path; and dynamically adjusting the operation force of the mechanical arm executing mechanism relative to the target object according to the force sense feedback data.
2. The control method according to claim 1, wherein the target object includes a target material frame and a target placement position, and the pose data includes a target gripping pose and a target placement pose; the visual perception unit is controlled to acquire pose data of the target object, and the method comprises the following steps: After the humanoid robot moves to a target operation area, controlling the visual perception unit to scan a target object to obtain original point cloud data comprising a target material frame and a target placement position; dividing the original point cloud data to separate a first point cloud cluster corresponding to the target material frame and a second point cloud cluster corresponding to the target placement position; Based on the first point cloud cluster, identifying a grabbing surface and grabbing points of the target material frame, and calculating to obtain a target grabbing pose of the target material frame under a robot base coordinate system; and calculating the target placement pose of the target placement position under the robot base coordinate system based on the second point cloud cluster.
3. The control method according to claim 1, wherein in driving the mechanical arm actuator to perform pose movement, controlling the environment sensing unit to scan a working area in real time and generate second environment point cloud data representing dynamic obstacle distribution, includes: driving the humanoid robot to move to the working area according to the static map; after the operation area is reached, controlling the environment sensing unit to perform three-dimensional scanning on the current operation area, and acquiring first environment point cloud data containing surrounding obstacles; And comparing and analyzing the first environmental point cloud data with the static map, removing background points and dynamic interference points overlapped with the static map, extracting barrier features which are changed or added relative to the static map, and generating the second environmental point cloud data.
4. The control method according to claim 1, wherein the determining whether to trigger an abnormal state based on the pose data and the force sense feedback data includes: Detecting the force sense feedback data in real time, and judging that collision or grabbing and slipping occur when the operation force mutation of the mechanical arm executing mechanism relative to the target object exceeds a preset force threshold value, and triggering a first abnormal state; and detecting the pose data in real time, and triggering a second abnormal state when detecting that the tracking characteristic of the target object is lost.
5. The control method of claim 1, wherein the robotic arm actuator comprises a first robotic arm and a second robotic arm, and wherein the unstacking strategy comprises: when the distance between the target material frame and the adjacent material frame is identified to be smaller than or equal to a safety threshold value, controlling the first mechanical arm to grasp a first side of the target material frame, dragging the target material frame to move to a preset separation distance along a direction away from the adjacent material frame, and controlling the second mechanical arm to grasp a second side of the target material frame, wherein the first side and the second side are opposite sides of the target material frame; and when the distance is confirmed to be larger than the safety threshold, controlling the first mechanical arm and the second mechanical arm to cooperate, and removing the target material frame from the pallet.
6. The control method of claim 1, wherein the robotic arm actuator comprises a first robotic arm and a second robotic arm, and wherein the palletizing strategy comprises: The method comprises the steps of acquiring and identifying a current operation scene according to pose data, determining a stacking target pose based on the pose of an empty pallet when the current operation scene is an initial empty pallet, and determining the stacking target pose based on the pose of a top-layer material frame when the current operation scene is an existing stack; When the stacking operation of the same layer of material frames is executed, calculating the expected distance between the target material frames and the placed material frames; when the expected distance is greater than or equal to a safety threshold, controlling a first mechanical arm and a second mechanical arm to cooperatively place the target material frame in a target pose and synchronously unloading the target material frame; When the expected distance is smaller than the safety threshold, the first mechanical arm and the second mechanical arm are controlled to cooperatively convey the target material frame to a preset position, the clamping jaw close to one side of the placed adjacent material frame is controlled to be released, the other side clamping jaw is controlled to independently hold the target material frame, and the mechanical arm on the holding side is controlled to move along the direction towards the placed adjacent material frame to push the target material frame until the target material frame is contacted with the placed adjacent material frame.
7. The control method of claim 4, wherein the exception handling policy comprises: When triggering the first abnormal state, controlling the mechanical arm executing mechanism to stop the current movement track and keep the current position in the preset time, controlling the mechanical arm executing mechanism to retract a preset safety distance along the abrupt change direction of the operation force, and after the retraction is completed, re-planning a path for grabbing or placing the target material frame and executing the compensation operation; When triggering the second abnormal state, controlling the mechanical arm executing mechanism to pause operation and move to a preset safe waiting position, if the target material frame is locked again in a preset time window, updating motion planning and restoring operation based on newly acquired position data, and if the target material frame is unlocked after overtime, triggering manual intervention alarm; and when the first abnormal state and the second abnormal state are both disappeared through real-time monitoring, controlling the mechanical arm executing mechanism to continuously execute subsequent actions along the motion track.
8. The control method according to claim 1, wherein correcting the motion trajectory in real time according to the second environmental point cloud data, to obtain an obstacle avoidance path, comprises: Constructing a local cost map containing dynamic obstacles in real time based on the second environmental point cloud data, wherein the dynamic obstacles comprise moving staff or target objects with positions changed; mapping the motion trail to the local cost map for collision detection; If the collision between the motion trail and the dynamic obstacle is detected, the joint angle sequence of the mechanical arm executing mechanism is planned online in the joint space, and the obstacle avoidance path is generated.
9. The control method according to claim 1, wherein dynamically adjusting the operating force of the mechanical arm actuator relative to the target object according to the force sense feedback data comprises: the method comprises the steps of controlling the mechanical arm executing mechanism to move to a preset rough positioning pose based on visual servo in a grabbing stage, switching to a constant force admittance control mode, calculating pose correction amount according to contact force deviation acquired by a force sense feedback unit in real time, and adjusting the tail end pose of the mechanical arm executing mechanism until reaching a precise positioning pose and completing grabbing the target object; And then switching to a constant force admittance control mode, calculating a pose correction amount and adjusting the tail end pose of the mechanical arm executing mechanism according to the contact force deviation acquired by the force sense feedback unit in real time, so that the target object approaches the target placing surface with constant contact force until the target object is placed.
10. A humanoid robot system for performing the method of any one of claims 1 to 9, characterized in that the system comprises: the environment sensing unit is used for scanning the working area and generating first environment point cloud data representing the static map; The visual perception unit is used for acquiring pose data of the target object; the system comprises a force sense feedback unit, a force sense control unit and a control unit, wherein the force sense feedback unit is used for detecting force sense feedback data of a mechanical arm executing mechanism in the process of contacting a target object; The system comprises a central control unit, an environment sensing unit, a robot shape robot, a robot shape map and a robot chassis, wherein the central control unit is used for judging whether an abnormal state is triggered according to the pose data and the force sense feedback data, identifying the arrangement type of a target material frame in the operation area according to the pose data when the abnormal state is not triggered, matching an unstacking strategy or a stacking strategy corresponding to the material frame arrangement type from a preset operation strategy library as a current execution strategy, selecting an abnormal processing strategy as the current execution strategy when the abnormal state is triggered, combining the space coordinates and the pose angle of the target material frame in the pose data and the first environment point cloud data, calculating an end effector path and an articulation sequence of the robot to generate a motion track of the robot, controlling the environment sensing unit to scan the operation area in real time and generate second environment point cloud data representing dynamic obstacle distribution in the process of driving the robot actuator to pose, and correcting the motion track in real time according to the second environment point cloud data to obtain the obstacle avoidance path, and controlling the robot shape map according to the static state to the static map to enable the robot shape map to move to the robot shape map to the robot chassis; The waist unit is connected between the chassis and the mechanical arm executing mechanism and used for driving the lifting and pitching angle of the mechanical arm executing mechanism; The manipulator actuating mechanism is used for continuously executing pose motion along the real-time obstacle avoidance path, dynamically adjusting the operation force of the manipulator actuating mechanism relative to the target object according to the force sense feedback data, calculating pose correction amount according to the force sense feedback data and adjusting the tail end pose of the manipulator actuating mechanism.
11. An electronic device comprising a memory for storing a computer program executable by the processor and a processor for executing the computer program in the memory to implement the method of any one of claims 1 to 9.
12. A computer readable storage medium having stored thereon a computer program, characterized in that the method according to any of claims 1 to 9 is enabled when the executable computer program in the storage medium is executed by a processor.
13. A computer program product comprising a computer program which, when executed by a processor, implements the method of any one of claims 1 to 9.

Description

Control method, system, device, medium, and program for humanoid robot Technical Field The present invention relates to the field of robot control technologies, and in particular, to a method, a system, an apparatus, a medium, and a program for controlling a humanoid robot. Background In the related art, the humanoid robot is used as an important carrier of personal intelligence, and has wide application prospect in unstructured scenes such as industrial assembly and logistics handling, e.g. unstacking and stacking tasks. In these industrial application scenarios, robots need to autonomously complete the full flow tasks from context awareness, path planning to final execution operations. However, existing control techniques still have significant limitations in practical applications. Firstly, in the aspect of motion planning and obstacle avoidance, the conventional method often models a static environment and detects a dynamic obstacle to perform a fracturing treatment. Most of the prior art only utilizes a laser radar or a depth camera to construct a static map once to generate an initial track in a task initialization stage, but lacks a continuous real-time scanning mechanism for the environment in the process of executing motion by a robot, or does not perform depth fusion and real-time correction on dynamic point cloud data generated in real time and the initial motion track although the dynamic point cloud data has dynamic detection capability. The control mode of open loop or semi-closed loop causes collision caused by delay perception or untimely re-planning when the robot faces suddenly appearing moving obstacles, and the motion safety under the complex dynamic environment is difficult to ensure. Secondly, in the interactive operation link of the mechanical arm and the target object, the prior art is mostly dependent on pure position control or simple visual servo positioning. Because the visual perception has errors, the mechanical arm has repeated positioning accuracy limitation, and the target object may have tiny pose deviation, the pure position control is extremely easy to generate rigid impact at the moment that the tail end of the mechanical arm contacts the object. Such impact forces may not only damage the delicate target object or the robotic end effector, but may also result in a gripping failure or object slip. The existing control scheme generally lacks a mechanism for collecting force sense feedback data in real time in the contact process and dynamically adjusting the operation force according to the force sense feedback data, so that 'soft and smooth' interaction similar to human beings cannot be realized, and the application capability of the humanoid robot in fine operations such as high-precision assembly or fragile article handling is limited. In the process of executing pose movement by the mechanical arm, a real-time judging and responding mechanism for abnormal states such as position deviation of a target object, sudden appearance of an environmental obstacle and blockage of the mechanical arm is often lacking, and operation failure or equipment damage is easily caused by sudden conditions. Accordingly, there is a need for a control method, system, apparatus, medium, and program for a humanoid robot to improve the above-described problems. Disclosure of Invention The invention provides a control method, a control system, control equipment, control media and control programs for a humanoid robot, which are used for solving the problems of safe operation and high-precision operation of the humanoid robot in a dynamic change environment. According to a first aspect of the embodiment of the invention, a control method of a humanoid robot is provided, which is applied to a humanoid robot system comprising a visual sensing unit, an environment sensing unit, a mechanical arm executing mechanism and a force sense feedback unit, and comprises the steps of controlling the environment sensing unit to scan a working area and generate first environment point cloud data representing a static map, controlling the visual sensing unit to acquire pose data of a target object, controlling the force sense feedback unit integrated at the tail end of the mechanical arm executing mechanism, detecting the force sense feedback data of the mechanical arm executing mechanism in the process of contacting the target object, judging whether an abnormal state is triggered according to the pose data and the force sense feedback data, identifying the arrangement type of the target material frame in a working area according to the pose data when the abnormal state is not triggered, matching a unstacking strategy or a stacking strategy corresponding to the arrangement type of the material frame from a preset working strategy base as a current executing strategy, selecting the abnormal processing strategy as the current executing strategy when the abnormal state is triggered, combining the space coordinates a