Method for adaptive estimation of spatial pose of robot arm by fusing motion information and visual information

Abstract

A kind of motion information and visual information fusion mechanical arm space pose self-adapting estimation method, first, the inner parameter matrix of binocular camera is combined, and three-dimensional reconstruction of two-dimensional image features is realized based on polar line constraint method;Then, the "hand-eye calibration" is completed by the combined measurement mode of laser tracker and binocular camera, and the relative position relationship between the machining end of mechanical arm and the center of workpiece is calculated based on the calibration relationship;Finally, an adaptive extended Kalman filter is proposed, the known binocular camera measurement noise covariance matrix is combined, the state noise covariance matrix of the end of mechanical arm at different positions is adaptively adjusted, the smooth estimation of the end position of mechanical arm during motion is realized, and finally the end of mechanical arm is accurately moved into the effective field of view range of binocular camera.The method solves the problem that the end position of mechanical arm is difficult to accurately estimate under complex working conditions, and the adaptive accurate estimation of the end position of mechanical arm can be realized without prior accurate observation and motion noise covariance matrix.

Description

Technical Field

[0001] This invention belongs to the fields of digital measurement and robotics, and relates to an adaptive estimation method for the spatial pose of a robotic arm that fuses motion information and visual information. Background Technology

[0002] In the aerospace field, high-end equipment components are becoming increasingly larger and more complex, gradually exceeding the operational range of machine tools. Mobile industrial robotic arms, with their advantages of large operating range, high flexibility, and high efficiency, hold great potential in processing such large parts. However, industrial robotic arms suffer from low absolute positioning accuracy, and relying solely on their inherent positioning accuracy is insufficient to meet processing precision requirements, necessitating the introduction of additional accuracy assurance technologies. Currently, methods to improve the absolute positioning accuracy of robotic arms mainly include offline calibration and online positioning. The former requires calibration and compensation of various joint parameters of the robotic arm to improve its absolute positioning accuracy; however, accurate calibration of joint parameters is difficult in complex on-site environments and when the end effector experiences dynamic changes with large loads. The latter uses additional sensors to acquire information from the end effector, and through online feedback of measurement information, it can accurately obtain the current actual pose of the end effector without the need for serial calculations of individual joint parameters, thus ensuring the absolute positioning accuracy of the end effector. This method holds great potential in robot-assisted positioning.

[0003] Among these methods, "eye-on-hand" robot end-effector assisted localization is one of the most widely used online localization methods. This involves mounting a binocular camera at the end of a robotic arm to acquire the coordinates of points in the target area at close range, and then transmitting this information to the robotic arm control system, thus achieving high-precision localization of the end-effector. However, binocular vision cameras have limited field of view and depth of field. In complex conditions, the end-effector often struggles to move directly to the binocular localization area, severely reducing the robustness and reliability of binocular localization. To reliably move the end-effector to the binocular visually measurable area, it is necessary to accurately estimate its spatial pose by combining real-time motion information. Traditional end-effector spatial pose estimation methods require accurate prior observations and motion state noise covariance matrices, which lack adaptability in complex conditions. Therefore, adaptive end-effector pose estimation methods suitable for complex environments are still needed.

[0004] Patent No. ZL202111045806.9, entitled "A Kinematic Self-Calibration Method for a Robotic Arm Based on Binocular Vision," discloses a calculation method for calculating the end-effector pose of a robotic arm by combining camera information and joint rotation angle information. Patent No. ZL201710936467.0, entitled "A Target Localization and Grasping Method for a Robotic Arm Based on Binocular Vision," discloses a method for estimating the grasping posture of a robotic arm based on binocular camera intrinsic parameters and image information. Both of these inventions achieve the estimation of the robotic arm pose through binocular cameras and other robotic arm parameters. However, during robotic arm operation, there are still challenges in accurately calibrating the joint parameters of the robotic arm, and the end-effector position cannot be accurately estimated based solely on motion information, making it impossible to accurately move the robotic arm within the effective field of view of the binocular camera. Therefore, this invention proposes an adaptive estimation method for the spatial pose of a robotic arm that fuses motion information and visual information, achieving accurate adaptive estimation of the end-effector pose. Summary of the Invention

[0005] This invention addresses the positioning needs of industrial robotic arms in operational scenarios. It tackles the problem of accurately estimating the end-effector pose information solely based on the robotic arm's own motion information, proposing an adaptive spatial pose estimation method that fuses motion and visual information. First, by combining the intrinsic parameter matrix of a binocular camera, three-dimensional reconstruction of image features is achieved using the epipolar constraint method. Then, "hand-eye calibration" is performed through a combination of laser tracker and binocular camera measurements. Based on the calibration relationship, the relative positional relationship between the robotic arm's end-effector and the workpiece center is calculated. Finally, an improved adaptive extended Kalman filter is proposed. Combined with the known binocular camera measurement noise covariance matrix, the state noise covariance matrix of the robotic arm's end-effector at different positions is adaptively adjusted, achieving a smooth estimation of the end-effector position during movement. Ultimately, this allows the robotic arm's end-effector to accurately move within the measurable range of the binocular camera, solving the problem of inaccurate end-effector position estimation under complex working conditions. This method balances accuracy and robustness, demonstrating good versatility and applicability.

[0006] The technical solution of the present invention:

[0007] An adaptive spatial pose estimation method for a robotic arm that fuses motion and visual information, comprising the following steps:

[0008] The first step is to obtain the 3D coordinates of feature points in the camera coordinate system.

[0009] First, the intrinsic and extrinsic parameters of the stereo cameras are calibrated to obtain the intrinsic parameter matrix K of the left and right cameras. L and K RAnd the extrinsic parameters R and t; then, visual markers are placed in the local area to be tested, and images of the area to be tested are acquired using a binocular camera. The images are then binarized to divide the foreground and background. The edges are then detected using an edge detection algorithm. Finally, the roundness constraint and area constraint are used to filter the image features. The result of the image filtering is shown in equation (1):

[0010]

[0011] Among them, (S) min ,S max T represents the set area threshold. C The set roundness threshold is defined as follows: C is the perimeter of each region's outline, S is the area of ​​each region's outline, and e represents the roundness of each region's outline.

[0012] Then, the fundamental matrix is ​​calculated based on the intrinsic parameter matrix of the stereo camera:

[0013]

[0014] in, The antisymmetric matrix obtained by calculating the translation vector between the left and right cameras;

[0015] Secondly, the center coordinates [u] of the visual landmarks in the feature regions of the camera-captured images are calculated using the ellipse fitting method. L v L 1] T and[u R v R 1] T And construct the following equation:

[0016]

[0017] Among them, [abc] T The basic matrix F and [u R v R 1] T The product;

[0018] Equation (3) can be transformed into a linear equation, which is the polar equation:

[0019] au L +bv L +c=0 (4)

[0020] When the distance between an image point and its corresponding epipolar line is less than the set constraint threshold, it is considered to be a matching point corresponding to an image point in another image.

[0021] Finally, by combining the depth information Z of each marker, the three-dimensional reconstruction of the marker is achieved:

[0022]

[0023] Where (XYZ) are the three-dimensional coordinates of the marker point in the camera coordinate system, c x and c y is the pixel coordinate of the camera's principal point, f is the camera's focal length, d is the parallax of the left and right images, and B is the distance between the optical centers of the left and right cameras.

[0024] The second step is online positioning of the robotic arm's end effector based on combined measurement.

[0025] First, place n laser trackers and visual cooperative target points on the calibration board, and measure their three-dimensional coordinates using both the laser trackers and binocular cameras:

[0026] P c ={p1 p2 p3 …p n}, Q l ={q1 q2 q3 …q n} (6)

[0027] Among them, P c and Q l These are the three-dimensional coordinates of the visual cooperative target point in the camera coordinate system and the laser tracker coordinate system, respectively.

[0028] Then, the transformation relationship between the camera coordinate system and the laser tracker coordinate system is solved based on the least squares principle:

[0029]

[0030] in, These are the rotation matrix and translation vector between the two coordinate systems, respectively.

[0031] Then, three laser tracker target mounts are installed on the fixture of the binocular camera, and their coordinates are measured using laser trackers. A fixture coordinate system is established using these three points, and the coordinates of these three points in the fixture coordinate system are calculated:

[0032]

[0033] Furthermore, the transformation relationship R between the tooling coordinate system and the laser tracker coordinate system is calculated. f l and t f l Based on the relationship between the camera coordinate system and the laser tracker coordinate system, the relationship R between the camera coordinate system and the tooling coordinate system is calculated. c f and t c f :

[0034]

[0035] Then, by changing the pose of the robotic arm, the transformation pose between the tooling coordinate system can be obtained under multiple robotic arm pose states. and pose transformation between the coordinate systems of the robotic arm's end effector This represents the pose transformation relationship of the tooling coordinate system from the j-th pose to the i-th pose. The pose transformation relationship of the robot arm's end effector coordinate system from the j-th pose to the i-th pose can be represented by the following relationship:

[0036]

[0037] in, and These are the relationships between the tooling coordinate system and the robotic arm end effector coordinate system for the j-th and i-th poses, respectively. This transformation relationship is fixed under any pose.

[0038] Right now:

[0039]

[0040] Where R and t represent the rotation and translation parts of the corresponding pose transformation relationship T, respectively;

[0041] This calculation process can be viewed as solving a typical nonlinear optimization problem AX = XB, obtained by solving the above system of equations. This refers to the transformation relationship between the tooling coordinate system and the robot arm end effector coordinate system. The transformation relationship R between the camera coordinate system and the robot arm end effector coordinate system can be calculated by combining formula (9). c e t c e This means completing the "hand-eye calibration":

[0042]

[0043] After establishing the "hand-eye alignment" relationship, a laser tracker target is fixed to the machining end of the flange at the end of the robotic arm. This target is used for positioning, and its coordinates are measured using a laser tracker. Based on the conversion relationship between the laser tracker and the binocular camera in equation (7) It can obtain the three-dimensional coordinates of the machining end positioning point in the camera coordinate system.

[0044]

[0045] Then, the three-dimensional coordinates of the center of the workpiece to be processed in the camera coordinate system are obtained according to equation (5). Based on the pose information obtained by the robotic arm itself, the transformation relationship between the robotic arm's end-effector coordinate system and the robotic arm's base coordinate system is solved. Solve for the three-dimensional coordinates of the machining end effector and the workpiece center in the robot arm's base coordinate system. and

[0046]

[0047] in, It is the transformation relationship between the camera coordinate system and the robot arm end effector coordinate system.

[0048] Finally, by calculating the relative position between the processing end and the workpiece center, the data is transmitted to the robotic arm controller to control the robotic arm end to reach the target area along the specified path.

[0049] The third step is to estimate the end effector position of the robotic arm based on an improved adaptive extended Kalman filter.

[0050] To enable the robotic arm to accurately reach the target area, an improved adaptive extended Kalman filter is used to dynamically estimate the end-effector position during the robotic arm's movement.

[0051] First, calculate the observation matrix of the system and linearize it:

[0052] Z k =H·X k +H·X k|k-1 +v k (15)

[0053] Among them, v k Let H be the system observation noise with a mean of 0 and a covariance matrix of R, and let X be the observation matrix. k X is the end-effector state vector at time k. k|k-1 It is a prediction of the state at time k based on information at time k-1;

[0054] Then, the motion state prediction equation of the robotic arm's end effector is:

[0055] X k =A·X k-1 +Bu k +w k (16)

[0056] Where A is the state transition matrix, B is the control input matrix, and u k w is the control vector k The motion state noise has a mean of 0 and a covariance matrix of Q.

[0057] Based on the initialized motion state noise covariance matrix Q, an adjustment ΔQ is introduced. The observation noise covariance matrix R can be obtained through calibration and is a known quantity. Therefore, the covariance matrix estimation error originates from ΔQ. The prediction equation for the actual error covariance matrix is ​​adjusted as follows:

[0058]

[0059] Among them, P k P is the error covariance matrix at time k. k|k-1 P is the error covariance matrix of the prediction at time k based on the information at time k-1. k-1 and Q k-1 The observation and motion state noise covariance matrices at time k-1 are ΔQ and ΔQ, respectively. k-1 It is the motion state noise covariance matrix adjusted at time k-1;

[0060] According to the covariance matching theory of Kalman filtering, the expected covariance of the observation residuals can be expressed as:

[0061]

[0062] in, It is the observation residual;

[0063] Then, based on the expected value of the observed residuals, the formula for adjusting the motion state noise covariance matrix is ​​derived:

[0064] ΔQ=(K·H) -1 (KE(ε k ·ε k T )K T -P k|k-1 H T ·K T )·((K·H) -1 ) T (19)

[0065] Where K is the Kalman gain;

[0066] Finally, based on the above information, the motion state noise covariance matrix is ​​updated. Motion state matrix X k Error covariance matrix:

[0067]

[0068] Process all observation data sequentially and dynamically adjust the motion state noise variance matrix. The system's state estimate is updated based on the new observations, enabling the robotic arm's processing end effector to move smoothly and accurately into the effective field of view of the binocular camera.

[0069] The beneficial effects of this invention are as follows: This invention proposes an improved adaptive Kalman filter, which, through a cyclic process of prediction-update-noise estimation-prediction, can achieve accurate dynamic estimation of the end position of the robotic arm without the need for prior accurate observation and motion state noise covariance matrix. It also ensures that the robotic arm system moves accurately, smoothly and stably into the binocular field of view, taking into account both accuracy and robustness. It solves the problem of difficulty in accurately estimating the end position of the robotic arm under complex working conditions and has broad application prospects. Attached Figure Description

[0070] Figure 1 This is a flowchart of a robotic arm spatial pose adaptive estimation method that integrates motion and visual information.

[0071] Figure 2 This is a scene diagram showing the pose estimation of the robotic arm's end effector during the processing of thin-walled cylindrical sections.

[0072] Figure 3 This is a schematic diagram of establishing the coordinate system of the binocular camera fixture.

[0073] Figure 4 It is the position estimation error of the robotic arm's end effector.

[0074] In the diagram: 1-robotic arm; 2-binocular camera; 3-camera fixture; 4-laser tracker; 5-calibration plate; 6-cooperative target point; 7-laser tracker target ball; 8-processing end effector; O B -X B Y B Z B O is the base coordinate system of the robotic arm. C -X C Y C Z C For the camera coordinate system, O L -X L Y L Z L Here is the coordinate system for the laser tracker. Detailed Implementation

[0075] The specific embodiments of the present invention will be described in detail below with reference to the technical solutions and accompanying drawings.

[0076] This embodiment uses an online positioning scenario in the process of simulating a robotic arm processing a simplified thin-walled cylinder segment, which includes 5 cooperative target points and 3 laser tracker target sphere points.

[0077] The first step is to obtain the 3D coordinates of feature points in the camera coordinate system.

[0078] First, the camera's intrinsic and extrinsic parameters are calibrated, resulting in the following intrinsic parameter matrix K. l K r And extrinsic parameter matrices R and t:

[0079]

[0080] Then, five cooperative target points were set on the calibration plate, and the coordinates of each target point in the camera coordinate system were obtained by using a binocular camera on the robotic arm:

[0081]

[0082] The second step is online positioning of the robotic arm's end effector based on combined measurement.

[0083] The coordinates of the cooperative target points are measured using a laser tracker. Based on the obtained three-dimensional coordinates of the marker points, the transformation relationship between the camera coordinate system and the laser tracker coordinate system is calculated. Then, three laser tracker target mounts are set on the fixture, and the coordinates of each point are measured using the laser tracker to establish the fixture coordinate system. Then, the relationship between the fixture coordinate system and the camera coordinate system is established. Then, the pose of the robotic arm is changed to obtain the pose transformation relationship between multiple fixture coordinate systems and the end-effector coordinate system of the robotic arm. Then, the "hand-eye calibration" is completed based on equations (10), (11), and (12). An electric spindle is installed at the end of the robotic arm, and a laser tracker target mount is fixed on the electric spindle as the positioning object. The coordinates of the target are measured using a laser tracker. Finally, the three-dimensional coordinates of the end of the model and the center of the workpiece in the base coordinate system of the robotic arm are calculated according to formulas (13) and (14).

[0084]

[0085] The third step is to estimate the end effector position of the robotic arm based on an improved adaptive extended Kalman filter.

[0086] Based on the above steps, the processing end needs to move from the initial position (281.84, 5.82, 834.19) to (281.84, -84.18, 744.19) along a straight line. Combining visual measurement information and robotic arm motion state information, an adaptive Kalman filter is used to achieve precise positioning of the end.

[0087] Set the initial parameters as follows:

[0088] The initial state variable X is:

[0089] X=[281.84,5.82,834.19,-38.4,-30.3] T

[0090] Initialize the state transition matrix A = I(3), and the input matrix B = [1,1,1];

[0091] The initial motion state noise covariance matrix is ​​as follows:

[0092] Q = 0.0001·I(3)

[0093] Initialize the observation noise covariance matrix:

[0094] R = 0.0025·I(15)

[0095] An improved adaptive Kalman filter dynamically adjusts the noise covariance matrix of the robotic arm's end effector motion, achieving accurate estimation of the end effector position. During the robotic arm's movement, a laser tracker uniformly measures the end effector position seven times. The measurement results are compared with the estimated end effector position, and the error is calculated, as shown in the attached figure. Figure 4 The effectiveness of the improved adaptive extended Kalman filter was verified.

[0096] This invention proposes an improved adaptive extended Kalman filter. Through a cyclic process of prediction-update-noise estimation-prediction, it can achieve accurate dynamic estimation of the end-effector position of a robotic arm without the need for prior accurate observation and motion state noise covariance matrix. It also ensures that the robotic arm system moves accurately, smoothly and stably into the field of view of the binocular camera, balancing accuracy and robustness. It solves the problem of difficult accurate estimation of the end-effector position under complex working conditions and has broad application prospects.

Claims

1. A method for adaptive estimation of spatial pose of a robotic arm by fusing motion information and visual information, characterized in that, The steps are as follows: The first step is to obtain the 3D coordinates of feature points in the camera coordinate system. First, the intrinsic and extrinsic parameters of the stereo cameras are calibrated to obtain the intrinsic parameter matrix K of the left and right cameras. L and K R And the extrinsic parameters R and t; then, visual markers are placed in the local area to be tested, and images of the area to be tested are acquired using a binocular camera. The images are then binarized to divide the foreground and background. The edges are then detected using an edge detection algorithm. Finally, the roundness constraint and area constraint are used to filter the image features. The result of the image filtering is shown in equation (1): Among them, (S) min ,S max T represents the set area threshold. C The set roundness threshold is defined as follows: C is the perimeter of each region's outline, S is the area of ​​each region's outline, and e represents the roundness of each region's outline. Then, the fundamental matrix is ​​calculated based on the intrinsic parameter matrix of the stereo camera: in, The antisymmetric matrix obtained by calculating the translation vector between the left and right cameras; Secondly, the center coordinates [u] of the visual landmarks in the feature regions of the camera-captured images are calculated using the ellipse fitting method. L v L 1] T and[u R v R 1] T And construct the following equation: Among them, [abc] T The basic matrix F and [u R v R 1] T The product; Equation (3) can be transformed into a linear equation, which is the polar equation: At L +bv L +c=0 (4) When the distance between an image point and its corresponding epipolar line is less than the set constraint threshold, it is considered to be a matching point corresponding to an image point in another image. Finally, by combining the depth information Z of each marker, the three-dimensional reconstruction of the marker is achieved: Where (XYZ) are the three-dimensional coordinates of the marker point in the camera coordinate system, c x and c y is the pixel coordinate of the camera's principal point, f is the camera's focal length, d is the parallax of the left and right images, and B is the distance between the optical centers of the left and right cameras. The second step is online positioning of the robotic arm's end effector based on combined measurement. First, place n laser trackers and visual cooperative target points on the calibration board, and measure their three-dimensional coordinates using both the laser trackers and binocular cameras: P c ={p1 p2 p3 …p n }, Q l ={q1 q2 q3 …q n } (6) Among them, P c and Q l These are the three-dimensional coordinates of the visual cooperative target point in the camera coordinate system and the laser tracker coordinate system, respectively. Then, the transformation relationship between the camera coordinate system and the laser tracker coordinate system is solved based on the least squares principle: in, These are the rotation matrix and translation vector between the two coordinate systems, respectively. Then, three laser tracker target mounts are installed on the fixture of the binocular camera, and their coordinates are measured using laser trackers. A fixture coordinate system is established using these three points, and the coordinates of these three points in the fixture coordinate system are calculated: Furthermore, the transformation relationship R between the tooling coordinate system and the laser tracker coordinate system is calculated. f l and t f l Based on the relationship between the camera coordinate system and the laser tracker coordinate system, the relationship R between the camera coordinate system and the tooling coordinate system is calculated. c f and t c f : R c f =(R f l ) -1 R c l ,t c f =(R f l ) -1 (t c l -t f l ) (9) Then, by changing the pose of the robotic arm, the transformation pose between the tooling coordinate system can be obtained under multiple robotic arm pose states. and pose transformation between the coordinate systems of the robotic arm's end effector This represents the pose transformation relationship of the tooling coordinate system from the j-th pose to the i-th pose. The pose transformation relationship of the robot arm's end effector coordinate system from the j-th pose to the i-th pose can be represented by the following relationship: in, and These are the relationships between the tooling coordinate system and the robotic arm end effector coordinate system for the j-th and i-th poses, respectively. This transformation relationship is fixed under any pose. Right now: Where R and t represent the rotation and translation parts of the corresponding pose transformation relationship T, respectively; This calculation process can be viewed as solving a typical nonlinear optimization problem AX = XB, obtained by solving the above system of equations. This refers to the transformation relationship between the tooling coordinate system and the robotic arm end effector coordinate system. The transformation relationship R between the camera coordinate system and the robot arm end effector coordinate system can be calculated by combining formula (9). c e t c e This means completing the "hand-eye calibration": After establishing the "hand-eye alignment" relationship, a laser tracker target is fixed to the machining end of the flange at the end of the robotic arm. This target is used for positioning, and its coordinates, P, are measured using a laser tracker. l m Based on the conversion relationship T between the laser tracker and the binocular camera in equation (7), l c It can obtain the three-dimensional coordinates of the machining end positioning point in the camera coordinate system. Then, the three-dimensional coordinates of the center of the workpiece to be processed in the camera coordinate system are obtained according to equation (5). Based on the pose information obtained by the robotic arm itself, the transformation relationship between the robotic arm's end-effector coordinate system and the robotic arm's base coordinate system is solved. Solve for the three-dimensional coordinates of the machining end effector and the workpiece center in the robot arm's base coordinate system. and in, It is the transformation relationship between the camera coordinate system and the robot arm end effector coordinate system. Finally, by calculating the relative position between the processing end and the workpiece center, the information is transmitted to the robotic arm controller to control the robotic arm end to reach the target area along the specified path; The third step is to estimate the end effector position of the robotic arm based on an improved adaptive extended Kalman filter. To enable the robotic arm to accurately reach the target area, an improved adaptive extended Kalman filter is used to dynamically estimate the end-effector position during the robotic arm's movement. First, calculate the observation matrix of the system and linearize it: Z k =H·X k +H·X k|k-1 +v k (15) Among them, v k Let H be the system observation noise with a mean of 0 and a covariance matrix of R, and let X be the observation matrix. k X is the end-effector state vector at time k. k|k-1 It is a prediction of the state at time k based on information at time k-1; Then, the motion state prediction equation of the robotic arm's end effector is: X k = A·X k-1 +Bu k +w k (16) Where A is the state transition matrix, B is the control input matrix, and u k w is the control vector k The motion state noise has a mean of 0 and a covariance matrix of Q. Based on the initialized motion state noise covariance matrix Q, an adjustment ΔQ is introduced. The observation noise covariance matrix R can be obtained through calibration and is a known quantity. Therefore, the covariance matrix estimation error originates from ΔQ. The prediction equation for the actual error covariance matrix is ​​adjusted as follows: Among them, P k P is the error covariance matrix at time k. k|k-1 P is the error covariance matrix of the prediction at time k based on the information at time k-1. k-1 and Q k-1 The observation and motion state noise covariance matrices at time k-1 are ΔQ and ΔQ, respectively. k-1 It is the motion state noise covariance matrix adjusted at time k-1; According to the covariance matching theory of Kalman filtering, the expected covariance of the observation residuals can be expressed as: in, It is the observation residual; Then, based on the expected value of the observed residuals, the formula for adjusting the motion state noise covariance matrix is ​​derived: ΔQ=(K H) -1 (KE(ε k ε k T )K T -P k|k-1 H T ·K T ) ((K H) -1 ) T (19) Where K is the Kalman gain; Finally, based on the above information, the motion state noise covariance matrix is ​​updated. Motion state matrix X k Error covariance matrix: Process all observation data sequentially and dynamically adjust the motion state noise variance matrix. The system's state estimate is updated based on the new observations, enabling the robotic arm's processing end effector to move smoothly and accurately into the effective field of view of the binocular camera.

Images