A flexible assembly robot with autonomous navigation and positioning capabilities

Abstract

The present invention relates to a flexible assembly robot with autonomous navigation and positioning functions, which includes a movable chassis and a control unit mounted on the chassis, a navigation device, a first binocular camera, and a Stewart parallel robot. The navigation device is composed of a laser radar, an inertial navigation system, and a second binocular camera. The second binocular camera is used to collect environmental point cloud data. The first binocular camera is used to collect a three-dimensional model and posture information of a workpiece to be assembled. The Stewart parallel robot is used to carry the workpiece to be assembled and is provided with a first pressure sensor for measuring the weight of the workpiece and a second pressure sensor for measuring the force of the assembly environment. The control unit is used to control the chassis and the Stewart parallel robot to perform corresponding actions based on the relevant data collected by the navigation device, the first binocular camera, the first pressure sensor, and the second pressure sensor. The present invention can effectively improve the working space range of the assembly robot and can reduce assembly stress or achieve stress-free assembly.

Description

Technical Field

[0001] The present invention relates to the field of robots, and in particular to a flexible assembly robot with autonomous navigation and positioning functions. Background Art

[0002] Docking and assembly are key processes in aerospace and marine equipment manufacturing. The production cycle and cost of this process significantly impact the overall product lifecycle. Flexible assembly robots with autonomous positioning capabilities are primarily used in automated assembly, primarily in the assembly of large, complex parts and numerous connectors in aircraft, ships, and automobiles.

[0003] The existing institutions have the following deficiencies:

[0004] 1) At present, the technologies used by most domestic assembly robots are mostly targeted at specific assembly objects, with low flexibility and flexibility, and they are immobile and have small working space;

[0005] 2) Currently, most domestic assembly robots cannot directly and effectively measure the impact of environmental forces on the stress of workpiece assembly and docking during the assembly process, and cannot eliminate the stress during the assembly process. Summary of the Invention

[0006] The present invention aims to provide a flexible assembly robot with autonomous navigation and positioning capabilities to solve the above problems. To this end, the specific technical solutions adopted by the present invention are as follows:

[0007] A flexible assembly robot with autonomous navigation and positioning functions is characterized in that it includes a movable chassis and a control unit, a navigation device, a first binocular camera and a Stewart parallel robot installed on the chassis, the navigation device is composed of a laser radar, an inertial navigation unit IMU and a second binocular camera, the second binocular camera is used to collect environmental point cloud data, the first binocular camera is used to collect the three-dimensional model and posture information of the workpiece to be assembled, the Stewart parallel robot is used to carry the workpiece to be assembled and is provided with a first pressure sensor for measuring the weight of the workpiece and a second pressure sensor for measuring the environmental force, the control unit is used to control the chassis and the Stewart parallel robot to perform corresponding actions in combination with the relevant data collected by the navigation device, the first binocular camera, the first pressure sensor and the second pressure sensor.

[0008] Furthermore, the chassis includes a frame and a power device and a drive device mounted on the frame. The power device is a rechargeable lithium battery. The drive device includes a Mecanum wheel, a motor driver, and a servo motor. The Mecanum wheel is fixed to the bottom of the frame. The servo motor is connected to the Mecanum wheel through the motor driver. The servo motor is powered by the rechargeable lithium battery.

[0009] Furthermore, the first binocular camera is mounted on the upper surface of the chassis via a camera bracket, and the second binocular camera is mounted on a side surface of the chassis.

[0010] Furthermore, the Stewart parallel robot includes an upper platform, a lower platform, six electric cylinders and a motion control card, wherein the upper platform is a dynamic platform for carrying the workpiece to be assembled, and the first pressure sensor is installed on the surface; the lower platform is a static platform, fixed on the upper surface of the chassis; the base and the upper end of the push rod of each electric cylinder are connected to a Hooke's hinge to form an independently moving motion unit, and the six motion units are connected to the upper and lower platforms by bolts, with a total of six degrees of freedom; the second pressure sensor is installed at the connection between the Hooke's hinge and the push rod of the electric cylinder; the electric cylinder is controlled by the motion control card, and the motion control card is electrically connected to the control unit.

[0011] Furthermore, the laser radar is installed on the lower platform.

[0012] Furthermore, the control unit includes a micro industrial computer and a touch screen display. The micro central control computer is installed in the chassis, and the touch screen display is installed on the side of the chassis and is electrically connected to the micro central control computer.

[0013] Furthermore, the microcomputer processes the data collected by the laser radar, the second binocular camera and the inertial navigation system, extracts image features using an improved ORB algorithm, optimizes by eliminating false matches through the PROSAC algorithm, fuses the binocular images of the second binocular camera, performs initial posture estimation, fuses the posture information obtained by IMU measurement to obtain a corrected rotation angle for posture optimization, and then uses the extended Kalman filter method to fuse the information of the second binocular camera and the laser radar to establish a computer-recognizable sparse point cloud environment map and a movable chassis model in the virtual space.

[0014] Furthermore, the micro-industrial computer obtains an image of the workpiece to be assembled through the first binocular camera, obtains three-dimensional point cloud features by performing three-dimensional reconstruction on the image, matches them with the CAD model features of the workpiece, and calculates the relative posture of the workpiece to be assembled using the Newton iteration method; then, the weight of the workpiece to be assembled is arranged on the first pressure sensor and its center of gravity is estimated using the least squares method, and the environmental force of the electric cylinder during the assembly process is measured through the second pressure sensor; combined with the artificial potential field theory and the kinematic model of the Stewart robot, the expected posture and operation trajectory of the assembly and docking of the workpiece to be assembled are calculated in a virtual environment.

[0015] The present invention adopts the above technical solution, which has the beneficial effects of: the present invention can effectively increase the working space range of the assembly robot, and can reduce assembly stress or achieve stress-free assembly. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] To further illustrate various embodiments, the present invention is provided with accompanying drawings. These drawings form part of the present disclosure and are primarily used to illustrate the embodiments and, in conjunction with the relevant description in the specification, to explain the operating principles of the embodiments. By referring to these drawings, one of ordinary skill in the art will understand other possible embodiments and the advantages of the present invention. The components in the figures are not drawn to scale, and similar reference numerals are generally used to represent similar components.

[0017] Figure 1 It is a three-dimensional diagram of a flexible assembly robot with autonomous navigation and positioning functions according to the present invention;

[0018] Figure 2 yes Figure 1 a perspective view of the chassis of the flexible assembly robot, wherein the upper surface is removed to illustrate the internal structure;

[0019] Figure 3 It is a flowchart of map construction and autonomous navigation;

[0020] Figure 4 It is a flow chart of workpiece 3D reconstruction and model matching;

[0021] Figure 5 This is the schematic diagram of the motion control of the Stewart parallel robot. DETAILED DESCRIPTION

[0022] The present invention will now be further described with reference to the accompanying drawings and specific embodiments.

[0023] like Figure 1 and 2As shown, a flexible assembly robot with autonomous positioning capabilities may include a movable chassis 1, a control unit mounted on the chassis 1, a navigation device, a first binocular camera 2, and a Stewart parallel robot 3. The movable chassis 1 is modified from the chassis of an AGV intelligent transport robot. The chassis 1 includes a frame, a power unit, and a drive unit mounted on the frame. The frame is welded from steel, with an aluminum alloy housing. The power unit utilizes a 48V rechargeable lithium battery 6. The drive unit comprises Mecanum wheels 11, a motor driver, and a servo motor. The control unit comprises a micro industrial computer 41 and a touch screen display 42 located in the center of the chassis 1. The touch screen 43 is connected to the micro industrial computer 41 and is used to display the robot's operating status and input control commands. The micro industrial computer 41 is connected to the motion control card and motor driver of the movable chassis 1. By controlling the forward and reverse rotation of the motor, the robot can achieve full-range motion, including forward and backward, longitudinal, lateral, diagonal, and rotational. The micro industrial computer 41 is connected to the motion control card 37 of the Stewart parallel robot 3 in the chassis 1 , and controls the posture of the Stewart parallel robot 3 by controlling the extension and contraction of each independent motion unit.

[0024] The navigation device consists of a second binocular camera 51 located on the top of the chassis 1, an inertial navigation unit 52 and a laser radar 53. The second binocular camera 51 is installed on the side (travel direction) of the chassis 1 to provide point cloud data of the assembly environment. The inertial navigation unit 52 is installed on the upper surface of the chassis 1 to provide data on angular velocity and acceleration. The laser radar 53 is installed on the lower platform 32 of the Stewart parallel robot 3 to obtain the distance to obstacles in the environment. The robot processes the relevant data collected by the second binocular camera 51, the inertial navigation unit 52 and the laser radar 53 through the micro industrial computer 41 to realize map construction and autonomous navigation. Specifically, as Figure 3As shown, the improved ORB algorithm is first used to extract features from the image captured by the second binocular camera 51. The PROSAC matching algorithm is used to eliminate false matches for optimization. The binocular images are then fused to obtain an initial pose. The pose measured by the inertial navigation unit (IMU) is used to correct the binocular camera's pose for pose optimization. The extended Kalman filter (EKF) method is combined with the lidar and binocular vision to fuse distance information and angle information respectively. A computer-recognizable sparse point cloud environment map and a movable chassis model are then established in virtual space. The IMU can correct for camera motion blur when the robot moves rapidly. The binocular camera can obtain rich texture and depth information, which can correct for drift in IMU readings when the robot is moving slowly. In other words, the IMU provides a better solution for fast movement, while the binocular camera can solve the IMU drift problem at slow speeds. Using IMU information to assist binocular vision can effectively combine the advantages of both sensors. LiDAR can provide precise depth information, but its point cloud density is too sparse, providing less than 6% of the total image points. This fails to cover all significant objects in the scene, and it lacks color information, which significantly hinders 3D scene understanding and perception. The combination of an IMU, a binocular camera, and LiDAR enables precise positioning and autonomous navigation.

[0025] The first binocular camera 2 is mounted on the chassis 1 via a bracket 21 to facilitate adjustment of its height and shooting direction, making it suitable for shooting different workpieces. The first binocular camera 2 is used to capture images of the workpiece to be assembled. The micro-industrial computer 41 performs three-dimensional reconstruction and model matching of the workpiece based on the images of the workpiece to be assembled captured by the first binocular camera 2, thereby obtaining the position information of the workpiece to be assembled. Specifically, Figure 4 As shown, the micro-industrial computer 41 calls the first binocular camera 2 to capture images of the workpiece and pre-processes the images: first, a series of images captured by the first binocular camera 2 are converted into grayscale images, and bilateral filtering is used to remove noise interference from each image. Then, the adaptive threshold Otsu method (OTUS) is used to binarize the entire image; then, the edge contour of the workpiece is extracted based on the Canny algorithm, and the point with the largest gradient direction in the image is obtained by the Sobel operator, that is, the point with the largest grayscale change is taken as the edge point, and the edge is extracted; then, the processed image is subjected to a three-dimensional reconstruction algorithm to obtain a three-dimensional point cloud model of the workpiece and extract the three-dimensional point cloud features, and then the workpiece CAD model is used as the reference model to extract the optimal matching point features of the CAD three-dimensional model, and finally, the pose of the workpiece is calculated by the Newton iteration method.

[0026] The Stewart parallel robot 3 consists of two upper and lower platforms 31 and 32. The upper platform 31 is a dynamic platform, used to carry the workpiece to be assembled, and a first pressure sensor 33 is installed on the surface. The first pressure sensor 33 is used to measure the weight of the workpiece; the lower platform 32 is a static platform, fixed to the upper surface of the movable chassis 1 by bolts; the base and the upper end of the push rod of each electric cylinder 34 are connected to a Hooke's hinge 35, forming a motion unit that can move independently. The connection between the Hooke's hinge 35 and the push rod of the electric cylinder 34 is equipped with a second pressure sensor 36. The second pressure sensor 36 is used to measure the environmental force. The six motion units are connected to the upper and lower platforms 31 and 32 by bolts, and have a total of six degrees of freedom. The motion control of the Stewart parallel robot is as follows: Figure 5 As shown, the specific process includes the following:

[0027] The first binocular camera 2 acquires an image of the workpiece to be assembled, performs 3D reconstruction, obtains 3D point cloud features, matches them with the CAD model features of the workpiece, and calculates the relative position and posture of the workpiece to be assembled using the Newton iteration method;

[0028] The weight of the workpiece to be assembled is measured by a first pressure sensor 33 arranged on the upper platform 31 of the Stewart parallel robot 3 and its center of gravity is estimated using the least squares method. At the same time, the environmental force acting on the electric cylinder during the assembly process is measured by a second pressure sensor 33;

[0029] Combining artificial potential field theory with the kinematic model of the Stewart robot, the desired position and trajectory of the assembly and docking of the workpiece to be assembled are calculated in a virtual environment.

[0030] The expansion and contraction amount of each electric cylinder of the Stewart platform is obtained through the inverse kinematic solution, and then the PID control is used to control each electric cylinder to move to the specified expansion and contraction amount in real time, so that the Stewart robot moves to the specified assembly posture.

[0031] The following briefly describes the working process of the present invention. Before assembly, the movable chassis first constructs a map of the assembly environment and uses a binocular camera to perform a three-dimensional reconstruction of the workpiece to be assembled. This is then matched with the workpiece's CAD model to calculate the workpiece's position and posture. When the operator issues an assembly instruction, the movable chassis first locates its position and, using a local path planning algorithm, selects an optimal transport route from its starting point to its destination on the map. Finally, a micro-industrial computer sends control commands to the motion control card, controlling the chassis to transport the workpiece along the planned route. Upon arrival at the designated assembly location, the micro-industrial computer calculates the relative position of the two workpieces using the Newton iteration method based on the image information captured by the binocular camera. The expected assembly path is then calculated in a virtual environment using artificial potential field theory and the dynamic model of a Stewart parallel robot. Based on the data collected by the binocular camera and pressure sensor, the micro-industrial computer controls the movement of the Stewart parallel robot to position the workpiece to the desired position. This involves using the Stewart parallel robot to adjust the workpiece's posture to achieve gravity compensation, thereby reducing assembly stress or achieving stress-free assembly.

[0032] The flexible assembly robot of the present invention has the following advantages over existing assembly robots:

[0033] 1) A movable chassis is added to the traditional Stewart parallel robot, which can map the assembly environment. When the starting and ending points of the assembly workpiece transportation are given, the robot can automatically plan the path and avoid obstacles, thereby increasing the working space of the assembly robot.

[0034] 2) The multi-sensor fusion technology is adopted, and the binocular camera is used to correct the drift of the IMU reading when the robot is moving slowly. The IMU can be used to correct the motion blur of the camera when the robot moves quickly. Combined with the advantages of lidar that it is less affected by light intensity, has a large measurement range and high sampling density, it has good robustness and calculation accuracy in measurement and positioning.

[0035] 3) Perform three-dimensional reconstruction of the workpiece, extract the three-dimensional point cloud features and match them with the workpiece CAD model, thereby improving the accuracy of workpiece reconstruction and positioning.

[0036] 4) By using digital assembly technology and combining it with the compliance control model of force / visual servoing, the assembly path of the workpiece can be calculated in a virtual environment based on the environmental forces and the relative position of the workpiece, which can reduce assembly stress or achieve stress-free assembly.

[0037] 5) The assembly robot is suitable for assembling one or more workpieces and can adapt to different assembly requirements with high flexibility and flexibility.

[0038] Although the present invention has been particularly shown and described in conjunction with preferred embodiments, it will be understood by those skilled in the art that various changes in form and details may be made to the present invention without departing from the spirit and scope of the invention as defined in the appended claims, and all such changes are within the scope of protection of the present invention.

Claims

1. A flexible assembly robot with autonomous navigation and positioning functions, characterized in that: The invention comprises a movable chassis and a control unit, a navigation device, a first binocular camera and a Stewart parallel robot installed on the chassis. The navigation device is composed of a laser radar, an inertial navigation unit (IMU) and a second binocular camera. The second binocular camera is used to collect environmental point cloud data. The first binocular camera is used to collect a three-dimensional model and posture information of a workpiece to be assembled. The Stewart parallel robot is used to carry the workpiece to be assembled and is provided with a first pressure sensor for measuring the weight of the workpiece and a second pressure sensor for measuring the environmental force. The control unit is used to control the chassis and the Stewart parallel robot to perform corresponding actions in combination with relevant data collected by the navigation device, the first binocular camera, the first pressure sensor and the second pressure sensor. The control unit includes a micro industrial computer and a touch display screen, the micro industrial computer is installed in the chassis, and the touch display screen is installed on the side of the chassis and is electrically connected to the micro industrial computer; the micro industrial computer obtains an image of the workpiece to be assembled through the first binocular camera, obtains three-dimensional point cloud features by performing three-dimensional reconstruction on the image, matches the three-dimensional point cloud features with the CAD model features of the workpiece, and calculates the relative posture of the workpiece to be assembled using the Newton iteration method; then, the weight of the workpiece to be assembled is measured through the first pressure sensor and its center of gravity is estimated using the least squares method, and the environmental force of the electric cylinder during the assembly process is measured through the second pressure sensor; and the expected posture and operation trajectory of the assembly and docking of the workpiece to be assembled are calculated in a virtual environment by combining the artificial potential field theory and the kinematic model of the Stewart robot.

2. The flexible assembly robot according to claim 1, characterized in that: The chassis includes a frame and a power device and a drive device mounted on the frame. The power device is a rechargeable lithium battery. The drive device includes a Mecanum wheel, a motor driver, and a servo motor. The Mecanum wheel is fixed to the bottom of the frame. The servo motor is connected to the Mecanum wheel through the motor driver. The servo motor is powered by the rechargeable lithium battery.

3. The flexible assembly robot according to claim 1, wherein: The first binocular camera is mounted on the upper surface of the chassis via a camera bracket, and the second binocular camera is mounted on a side surface of the chassis.

4. The flexible assembly robot according to claim 1, wherein: The Stewart parallel robot includes an upper platform, a lower platform, six electric cylinders and a motion control card, wherein the upper platform is a dynamic platform for carrying the workpiece to be assembled, and the first pressure sensor is installed on the surface; the lower platform is a static platform, fixed to the upper surface of the chassis; the base and the upper end of the push rod of each electric cylinder are connected to a Hooke's hinge to form an independently moving motion unit, and the six motion units are connected to the upper and lower platforms by bolts, with a total of six degrees of freedom; the second pressure sensor is installed at the connection between the Hooke's hinge and the push rod of the electric cylinder; the electric cylinder is controlled by the motion control card, and the motion control card is electrically connected to the control unit.

5. The flexible assembly robot according to claim 4, characterized in that: The laser radar is installed on the lower platform.

6. The flexible assembly robot according to claim 1, characterized in that: The micro-industrial computer processes the data collected by the laser radar, the second binocular camera and the inertial navigation system, extracts image features using an improved ORB algorithm, optimizes by eliminating false matches using the PROSAC algorithm, fuses the binocular images of the second binocular camera, performs initial pose estimation, fuses the pose information obtained by IMU measurement to obtain a corrected rotation angle for pose optimization, and then uses an extended Kalman filter method to fuse the information of the second binocular camera and the laser radar to establish a computer-recognizable sparse point cloud environment map and a movable chassis model in a virtual space.

Images