A humanoid robot passable area generation method and device and a humanoid robot

Abstract

The application belongs to the technical field of robots, and provides a humanoid robot passable area generation method and device and a humanoid robot, the method comprising: in the case of detecting that the humanoid robot is powered on and starts to move, acquiring an original image of a to-be-passed area of the humanoid robot; performing defocus recovery processing on the original image, and generating a depth image based on the distance from each pixel point in the defocus-recovered image to a monocular camera; determining a first area composed of pixel points with a depth value greater than a distance threshold in the depth image, and mapping the first area to the original image to obtain a second area; determining an impassable area of the original image; determining a passable area of the humanoid robot based on the impassable area of the original image; determining a passing direction angle of the humanoid robot, and controlling the humanoid robot to pass in the passable area based on the passing direction angle. The application can accurately generate the passable area of the humanoid robot in real time, and perform real-time navigation.

Description

Technical Field

[0001] This invention belongs to the field of robotics technology, and specifically relates to a method, apparatus, and humanoid robot for generating a passable area for a humanoid robot. Background Technology

[0002] Humanoid robots are robots designed to mimic human appearance and behavior. Because humanoid robots can assist people in completing relatively heavy and complex tasks, and because humanoid robots have high work efficiency and quality, they are widely used in many industries such as medicine, military, education, and daily life.

[0003] Current humanoid robots rely excessively on hardware, such as LiDAR, binocular cameras, or structured light sensors, in generating traversable areas. Low image resolution from the hardware leads to low depth estimation accuracy, which in turn results in low accuracy of the generated traversable area. Furthermore, the use of traditional image processing algorithms in existing humanoid robots results in poor real-time performance of their vision systems, meaning they cannot accurately generate traversable areas in real time. Summary of the Invention

[0004] The purpose of this invention is to generate passable areas for humanoid robots in real time and accurately, and to ensure that humanoid robots can still work stably in complex environments and scenarios such as complex lighting conditions.

[0005] In a first aspect, embodiments of the present invention provide a method for generating a passable area for a humanoid robot, applicable to a humanoid robot including a monocular camera, the method comprising:

[0006] When the humanoid robot is detected to be powered on and starting to move, the original image of the area to be traversed by the humanoid robot is acquired, and the original image is acquired in real time by the monocular camera;

[0007] The original image is defocused and restored to obtain a defocused image, and a depth image is generated based on the distance from each pixel in the defocused image to the monocular camera.

[0008] A first region is determined, consisting of pixels in the depth image whose depth values ​​are greater than a distance threshold, and the first region is mapped onto the original image to obtain a second region; the distance threshold is determined based on the current travel speed of the humanoid robot and a preset time.

[0009] The impassable region of the original image is determined as a rectangular region based on the width of the second region and the height of the original image.

[0010] Based on the impassable areas in the original image, the traversable areas of the humanoid robot are determined in the original image;

[0011] The travel direction angle of the humanoid robot is determined, and based on the travel direction angle, the humanoid robot is controlled to travel in the passable area.

[0012] Optionally, determining the first region composed of pixels in the depth image whose depth values ​​are greater than a distance threshold includes:

[0013] The current travel speed of the humanoid robot is obtained, and the product of the current travel speed and the preset time is determined as the distance threshold;

[0014] In the depth image, locate target pixels whose distance is greater than the distance threshold, and determine the width of the region formed by the target pixels;

[0015] Based on the width of the region formed by the target pixels and the height of the depth image, a first region is determined, which is composed of pixels in the depth image whose depth value is greater than a distance threshold.

[0016] Optionally, determining the passable area of ​​the humanoid robot in the original image based on the impassable area of ​​the original image includes:

[0017] Calculate the width of each rectangular region in the original image, excluding the impassable region;

[0018] The width of the humanoid robot is obtained, and the product of the width of the humanoid robot and a preset multiple is determined as the width threshold.

[0019] The width of each rectangular region is compared with the width threshold.

[0020] If there is a target rectangular region with a width greater than the width threshold, the rectangular region with the largest width in the target rectangular region is determined as the passable area of ​​the humanoid robot.

[0021] Optionally, determining the travel direction angle of the humanoid robot and controlling the humanoid robot to travel within the passable area based on the travel direction angle includes:

[0022] Determine the coordinates of the first center pixel of the passable area, and determine the coordinates of the second center pixel of the original image;

[0023] Determine the depth value of the first center pixel, wherein the depth value of the first center pixel is the depth value of the center pixel of the depth image;

[0024] Based on the coordinates of the first center pixel, the coordinates of the second center pixel, and the depth value of the first center pixel, the travel direction angle of the humanoid robot is calculated;

[0025] According to the humanoid robot's travel direction angle and current travel speed, control the humanoid robot to travel in the passable area, and return to execute the step of acquiring the original image of the humanoid robot's area to be traveled.

[0026] Optionally, the method further includes:

[0027] If there is no passable area for the humanoid robot in the original image, control the humanoid robot to rotate in place by a preset angle and control the humanoid robot's passing speed to zero;

[0028] Return to the step of obtaining the original image of the area to be traversed by the humanoid robot.

[0029] Optionally, the step of performing defocus restoration processing on the original image to obtain a defocus-restored image includes:

[0030] The original image is input into a pre-trained defocused image restoration model to perform defocus restoration processing on the original image and output the defocused image restored.

[0031] The defocused image restoration model is obtained by training with pre-acquired defocused images and corresponding sharp images.

[0032] Optionally, generating a depth image based on the distance from each pixel in the defocused image to the monocular camera includes:

[0033] The image after defocus restoration is input into a pre-trained image depth prediction model so that the image depth prediction model extracts the image features of the image after defocus restoration. Based on the relationship between image features and depth distance, the distance from each pixel in the image after defocus restoration to the monocular camera is obtained, and a depth image with the distance as the depth is output.

[0034] The image depth prediction model is trained using clear images from an image library.

[0035] Secondly, embodiments of the present invention also provide a humanoid robot traversable area generation device, applied to a humanoid robot including a monocular camera, the device comprising:

[0036] The original image acquisition module is used to acquire the original image of the area to be traversed by the humanoid robot when the robot is detected to be powered on and starting to move. The original image is acquired in real time by the monocular camera.

[0037] The depth image generation module is used to perform defocus restoration processing on the original image to obtain a defocus restored image, and generate a depth image based on the distance from each pixel in the defocus restored image to the monocular camera.

[0038] The region determination module is used to determine a first region composed of pixels in the depth image whose depth values ​​are greater than a distance threshold, and to map the first region onto the original image to obtain a second region; the distance threshold is determined based on the current travel speed of the humanoid robot and a preset time.

[0039] An impassable region determination module is used to determine impassable regions of the original image, wherein the impassable region is a rectangular region determined based on the width of the second region and the height of the original image;

[0040] The passable area determination module is used to determine the passable area of ​​the humanoid robot in the original image based on the impassable areas in the original image.

[0041] The humanoid robot control module is used to determine the travel direction angle of the humanoid robot and, based on the travel direction angle, control the humanoid robot to travel in the passable area.

[0042] Thirdly, embodiments of the present invention also provide a humanoid robot, comprising:

[0043] A monocular camera is used to capture raw images of the area to be traversed by the humanoid robot in real time.

[0044] At least one processor is configured to acquire the raw image from the monocular camera upon detecting that the humanoid robot has been powered on and begun to move;

[0045] Memory for storing the at least one processor-executable instruction;

[0046] The at least one processor is configured to execute the instructions to implement the method described in the first aspect.

[0047] Fourthly, embodiments of the present invention provide a computer-readable storage medium that, when instructions in the computer-readable storage medium are executed by a processor of an electronic device, enables the electronic device to perform the method described in the first aspect.

[0048] The technical solution provided in this invention uses a humanoid robot's monocular camera to acquire original images of the area to be traversed in real time. Since the original image may be a low-resolution, out-of-focus image, defocus restoration processing is performed to obtain a defocus-restored image. A depth image is then generated based on the distance from each pixel in the defocus-restored image to the monocular camera. Because the defocus-restored image has higher resolution, the accuracy of the distances from each pixel to the monocular camera is also higher, resulting in a more accurate depth image; that is, the depth values ​​of each pixel in the generated depth image are highly accurate.

[0049] Then, a first region is determined, consisting of pixels with depth values ​​greater than a distance threshold in the depth image. Since this region is far from the humanoid robot, it will not move to such a distant region. Therefore, the first region is considered an impassable region determined by the depth image. Furthermore, to ensure that the final traversable region corresponds to the actual spatial region, the first region in the depth image is mapped onto the original image to obtain a second region corresponding to the first region. Since the humanoid robot moves vertically, the rectangular region defined by the width of the second region and the height of the original image is determined as the impassable region of the original image. Based on this impassable region, the traversable region of the humanoid robot is determined in the original image. Finally, the travel direction angle of the humanoid robot is determined, and the robot is controlled to move within the traversable region according to this angle.

[0050] As can be seen, the technical solution of this invention can accurately generate the passable area of ​​the humanoid robot in real time, accurately determine the passing direction angle in real time, and control the humanoid robot to pass within the passable area in real time according to the passing direction angle, thereby realizing real-time navigation. Furthermore, the humanoid robot can still work stably in complex environments and scenarios such as complex lighting conditions, enhancing its robustness. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of a portion of the hardware structure of the humanoid robot provided in an embodiment of the present invention;

[0052] Figure 2 A flowchart illustrating the overall technical solution of a method for generating a passable area for a humanoid robot, as provided in an embodiment of the present invention;

[0053] Figure 3 A flowchart illustrating a method for generating a passable area for a humanoid robot, provided in an embodiment of the present invention;

[0054] Figure 4 A flowchart illustrating a specific implementation of S360 provided in this embodiment of the invention;

[0055] Figure 5 A schematic diagram of a humanoid robot traversable area generation device provided in an embodiment of the present invention;

[0056] Figure 6 A schematic diagram of the structure of a humanoid robot provided in an embodiment of the present invention. Detailed Implementation

[0057] The present invention will be described in detail below through embodiments.

[0058] With the rapid development of science and technology, humanoid robots are widely used in various industries such as medicine, military, education, and daily life. Existing methods for generating navigable areas and navigating humanoid robots rely excessively on hardware, such as LiDAR, binocular cameras, or structured light sensors. This not only increases the weight of the humanoid robot's head but also hinders the improvement of its visual resolution, and the overall cost of the humanoid robot is relatively high.

[0059] Furthermore, existing humanoid robot navigation systems do not integrate image restoration processing functions. For example, existing humanoid robots do not have image defocus restoration functions, which makes the system unable to work effectively when faced with blurry or defocused images. This greatly reduces the accuracy of the extracted depth information, resulting in low accuracy of the generated humanoid robot's traversable area and problems with the humanoid robot's path planning.

[0060] Meanwhile, existing humanoid robots, which use traditional image processing algorithms, suffer from insufficient real-time performance in their vision systems. This prevents them from meeting the rapid response requirements of humanoid robots in dynamic environments; in other words, they cannot accurately generate navigable areas in real time. Consequently, existing humanoid robots rely on supplementary sensors such as six-dimensional force sensors, which may further hinder their ability to generate navigable areas promptly and accurately in emergency situations, thus preventing accurate navigation.

[0061] In summary, existing humanoid robots do not effectively integrate image restoration, depth estimation, and path planning, resulting in low efficiency and low accuracy in generating traversable areas. This leads to problems such as slow response and maintenance difficulties in existing humanoid robots.

[0062] To address the aforementioned problems with existing humanoid robots, this invention provides a method, apparatus, and humanoid robot for generating passable areas for humanoid robots.

[0063] The humanoid robot passage area generation method provided by this invention combines monocular image defocus restoration technology and depth estimation technology to generate passage areas for the humanoid robot in real time. This method achieves high efficiency and accuracy in generating passage areas. Furthermore, even in complex environments or low-light conditions, the humanoid robot can efficiently and accurately generate passage areas. In other words, the humanoid robot provided by this invention can quickly respond to environmental changes and provide accurate spatial information for path planning, overcoming the limitations of existing technologies in humanoid robot navigation and path planning, and improving the humanoid robot's environmental perception, navigation accuracy, and path planning efficiency. Moreover, by using a monocular camera, the weight of the humanoid robot's head can be reduced, saving costs.

[0064] To ensure clarity, the overall technical solution of this invention will be described in detail below in conjunction with specific application scenarios.

[0065] First, the hardware structure of the humanoid robot provided in the embodiments of the present invention will be introduced.

[0066] like Figure 1 As shown, the hardware structure of the humanoid robot provided in this embodiment of the invention includes: a humanoid robot head vision mounting base 101, a communication cable routing structure 102, and a monocular camera 103, all of which are mounted on a flat plate on the chest of the humanoid robot.

[0067] The humanoid robot head vision mounting base 101 is used for overall installation and locking, and is compatible with the upper surface of the chest of various types of humanoid robots. The communication cable routing structure 102 is used to house the communication cable for neat cable management. The monocular camera 103 is a general-purpose monocular area array industrial camera of various types, with no special requirements for specifications or models. It is used to acquire real-time monocular images of the area to be traversed by the humanoid robot during its movement, so that in subsequent steps, the traversable area of ​​the humanoid robot can be efficiently and accurately generated using these monocular images.

[0068] It should be explained that when securing the monocular camera 103 to the humanoid robot's head vision mounting base 101, the installation only needs to be secure enough to prevent the monocular camera 103 from falling off; there are no special calibration steps or installation angle requirements. The camera communication cable is then embedded in the communication cable routing structure 102; the embedding method has no special requirements. Therefore, the solution of this embodiment of the invention has the advantage of simple installation.

[0069] After describing the hardware structure of the humanoid robot provided in the embodiments of the present invention, the overall technical solution of the method for generating a passable area for a humanoid robot provided in the embodiments of the present invention will be described below. The executing entity of this method for generating a passable area for a humanoid robot can be... Figure 1The humanoid robot described in the embodiment is equipped with a humanoid robot traversable area generation system for performing tasks such as... Figure 2 The method for generating traversable areas for the humanoid robot is shown. For example... Figure 2 As shown, the method for generating a passable area for the humanoid robot may include the following steps:

[0070] S210: Upon detecting that the humanoid robot has been powered on and has begun to move, the monocular camera is activated.

[0071] Specifically, after detecting that the humanoid robot has been powered on and started moving, it is necessary to generate a passable area for the robot in real time so that the robot can be controlled to move within the generated passable area. Therefore, a monocular camera needs to be activated so that images of the area to be traversed can be acquired in subsequent steps.

[0072] S220, to determine if the monocular camera communication is normal.

[0073] Specifically, after starting the monocular camera via S210, to ensure that the monocular camera can acquire images in real time, it is necessary to determine whether the monocular camera communication is normal. If the monocular camera communication is normal, it means that the monocular camera can work normally, that is, it can acquire images in real time. At this time, proceed to the next step S230. If the monocular camera communication is abnormal, it means that the monocular camera cannot work normally, that is, it cannot acquire images. At this time, continue to execute S220 until it is determined that the monocular camera communication is normal.

[0074] S230, under the condition that the monocular camera is communicating normally, acquires the original image captured by the monocular camera in real time, calls the defocus restoration algorithm to restore the sharpness of the original image, and obtains the defocus restored image.

[0075] Specifically, assuming the monocular camera is communicating normally, it can capture images of the area to be traversed in real time. For clarity, these images can be referred to as the raw images. Since the raw images directly captured by the monocular camera may be out-of-focus and lack sharpness, a defocus restoration algorithm can be used to restore the original image's sharpness in subsequent steps to accurately generate the traversable area for the humanoid robot. This results in a defocus-restored image.

[0076] In this embodiment, the specific implementation of calling the defocus restoration algorithm to perform defocus restoration processing on the original image can be as follows: the original image is directly input into the pre-trained defocus image restoration model, the defocus image restoration model performs defocus restoration processing on the defocus image, and outputs the defocus restored image.

[0077] The training process for the defocused image restoration model can be as follows: Collect a large number of defocused images of various types as negative samples; collect a large number of sharp images that have a mapping relationship with the defocused images in the negative samples as positive samples; input the negative and positive samples into the neural network model to be trained; when the accuracy of the sharp images output by the neural network model after inputting the negative samples is greater than a preset accuracy, it indicates that the trained neural network model can restore the defocused image to a sharp image, and at this point, the trained neural network model is determined as the defocused image restoration model. The preset accuracy can be determined according to the actual situation and is not specifically limited here.

[0078] As can be seen, the technical solution of this invention does not perform preprocessing or image enhancement on the original image during the process of restoring the sharpness of the defocused image. Instead, it directly inputs the original image into the pre-trained defocused image restoration model, which can efficiently obtain a clear image after defocus restoration, and the accuracy of the obtained clear image is high.

[0079] S240, call the depth estimation algorithm to calculate the depth of the defocused image, obtain the distance of each pixel in the defocused image to the monocular camera, and form a depth image with the distance as the depth.

[0080] Specifically, after obtaining the defocused image, a depth estimation algorithm can be used to calculate the depth of the defocused image, obtaining the distance from each pixel in the defocused image to the monocular camera, and forming a depth image with that distance as the depth. Because the defocused image has high clarity, the accuracy of the distances from each pixel to the monocular camera obtained by performing depth calculations on the defocused image is high, resulting in a highly accurate depth image. This helps in accurately determining the passable area of ​​the humanoid robot in subsequent steps.

[0081] In this embodiment, the specific implementation of S240 can be as follows: the image after defocus restoration is input into a pre-trained image depth prediction model, the image depth prediction model extracts the image features of the image after defocus restoration, and based on the relationship between the pre-learned image features and the depth distance, the depth distance of each pixel in the image after defocus restoration is obtained, and a depth image is generated based on the depth distance.

[0082] In practical applications, the process of training an image depth prediction model can be as follows: A large number of super-resolution images from an image library are input into the image depth prediction model to be trained. The model is then trained, extracting image features from the super-resolution images and determining the depth distance of each pixel based on these features. If the accuracy of the pixel depth distance output by the image depth prediction model is greater than a preset accuracy, it indicates that the trained image depth prediction model can accurately predict the depth distance of each pixel in the image. In this case, the trained image depth prediction model is determined as the final image depth prediction model. The preset accuracy can be determined based on actual conditions and is not specifically limited here.

[0083] As can be seen, the technical solution of this invention, when calculating the distance from each pixel in the defocused image to the monocular camera, inputs the defocused image to a pre-trained defocused image restoration model, thereby efficiently and accurately obtaining the distance from each pixel in the defocused image to the monocular camera.

[0084] It is understandable that there is a one-to-one mapping between the pixels of the defocused image and the pixels of the original image, and similarly, there is a one-to-one mapping between the pixels of the depth image and the pixels of the defocused image, and consequently, there is also a one-to-one mapping between the pixels of the depth image and the pixels of the original image. After obtaining the distance from each pixel of the depth image to the monocular camera, by mapping each pixel of the depth image to the pixels of the original image, the distance from each pixel of the original image to the monocular camera can be accurately obtained.

[0085] S250, determine the first region in the depth image where the depth value of a pixel is greater than a distance threshold.

[0086] The distance threshold is determined by multiplying the humanoid robot's current travel speed by a preset duration. Assuming the preset duration is 5 seconds, the distance threshold is the current travel speed multiplied by 5 seconds. Of course, the specific value of the preset duration can also be determined based on the actual duration, and this embodiment of the invention does not impose a specific limitation on it. Furthermore, when the humanoid robot first starts, its current running speed is a preset initial running speed. This initial running speed can also be determined based on the actual situation, and this embodiment of the invention does not impose a specific limitation on it either.

[0087] Since areas with a distance greater than a distance threshold are far from the humanoid robot's current position, the robot will not move to these distant areas. Therefore, to improve the robot's response speed in determining traversable areas, a first region with pixel depth values ​​greater than the distance threshold can be identified in the depth image. This first region is the traversable region determined by the depth image. In practical applications, this traversable region is a rectangular area. The width of this rectangular region is the width of the area formed by pixels with depth values ​​greater than the distance threshold, and the height of this rectangular region can be the height of the depth image. There can be one or more traversable regions in the depth image.

[0088] S260, based on the mapping relationship between the pixels of the depth image and the pixels of the original image, each first region obtained in step S250 is mapped to the original image to obtain a second region corresponding to the first region; the first rectangular region is determined by using the height of the original image as the rectangle height and the width of each second region as the rectangle width; and the first rectangular region is subtracted from all regions of the original image to obtain the second rectangular region.

[0089] Specifically, in order to ensure that the finally determined passable areas correspond to the actual spatial areas, after marking the impassable areas in the depth image, it is also necessary to map the width of the impassable areas in the depth image to the original image. That is, the first area is mapped to the original image to obtain the second area, and the width of the second area is the width of the impassable area in the original image.

[0090] In practical applications, the size of the depth image can be the same as or different from the size of the original image. If the size of the depth image is the same as the size of the original image, the width of the second region in the mapped original image is the same as the width of the first region obtained in step S250. If the size of the depth image is different from the size of the original image, the width of the second region in the mapped original image is the same as the width of the first region obtained in step S250. Since the humanoid robot moves vertically, after obtaining the width of each second region in the original image, a first rectangular region is formed in the original image using the width of each second region as the rectangle width and the height of the original image as the rectangle height. This first rectangular region is the impassable region in the original image. Then, the first rectangular region is subtracted from the total area of ​​the original image to obtain the second rectangular region. In practical applications, the number of second rectangular regions can be one or more.

[0091] S270: Obtain the width of the humanoid robot, compare the width of the second rectangular region obtained in step S260 with the width threshold, mark the rectangular regions whose width in step S260 is less than the width threshold as new impassable regions, and add them to the set of impassable regions marked in step S260.

[0092] The width threshold is determined by the width of the humanoid robot and a preset multiple. For example, the preset multiple could be 2, in which case the width threshold is the width of the humanoid robot multiplied by 2. Of course, in practical applications, the specific value of the preset multiple can be determined according to the actual situation, as long as it ensures that the humanoid robot can safely pass through the actual space corresponding to the passable rectangular area.

[0093] If the width of a certain second rectangular region determined in step S260 is less than the width threshold, it means that when the humanoid robot passes through the actual space area corresponding to the second rectangular region, a collision may occur, that is, it may be unable to pass. Therefore, in order to ensure that the humanoid robot can pass safely, the second rectangular region with the smaller width needs to be determined as an impassable region and added to the set of impassable rectangular regions marked in step S260.

[0094] S280, determine whether there is a third rectangular region with a width greater than the width threshold in the second rectangular region excluding the impassable region in the original image.

[0095] Specifically, through steps S260 and S270, all impassable rectangular regions in the original image can be determined. In order to generate the passable region for the humanoid robot, it is necessary to determine whether there is a third rectangular region with a width greater than the width threshold in the second rectangular region excluding the impassable region in the original image.

[0096] S290, if there is no third rectangular region in the original image, the humanoid robot rotates in place by a preset angle, sets its travel speed to 0, and returns to step S230.

[0097] Specifically, if the original image contains only impassable areas, meaning the number of remaining third rectangular areas in the original image is 0, then the humanoid robot currently has no passable areas. In this case, the humanoid robot can be controlled to rotate in place by a preset angle. This preset angle can be set according to the actual situation and is not specifically limited here. For example, the preset angle can be +45°, indicating that the humanoid robot can turn 45° clockwise in place, and the preset angle can be -45°, indicating that the humanoid robot can turn 45° counterclockwise in place. After controlling the humanoid robot to rotate in place by the preset angle, return to step S220, that is, regenerate the passable areas of the humanoid robot.

[0098] S2100, if a third rectangular region exists in the original image, adjust the humanoid robot's travel angle, and control the humanoid robot to travel in the space corresponding to the third rectangular region according to the adjusted travel angle and the current travel speed, and return to execute step S320.

[0099] Specifically, if the number of remaining third rectangular regions in the original image is greater than 1, it indicates that a passable area exists in the original image. In this case, the rectangular region with the largest width among all third rectangular regions is selected as the passable area. The coordinates (x, y) of the center point of this passable area are calculated, and the distance d from the center point of this passable area to the monocular camera is obtained in the depth image. The coordinates (x0, y0) of the center point of the original image are also obtained. Arcsin(x-x0) / d is used as the travel direction angle of the humanoid robot, while keeping the current travel speed constant. The humanoid robot is controlled to travel within the passable area with the largest width according to the travel direction angle and the current travel speed, and then the process returns to step S220. This allows for real-time determination of the humanoid robot's passable area and control of the humanoid robot to travel within the passable area, thereby achieving real-time navigation.

[0100] The technical solution provided in this invention uses a humanoid robot's monocular camera to acquire original images of the area to be traversed in real time. Since the original image may be a low-resolution, out-of-focus image, defocus restoration processing is performed to obtain a defocus-restored image. A depth image is then generated based on the distance from each pixel in the defocus-restored image to the monocular camera. Because the defocus-restored image has higher resolution, the accuracy of the distances from each pixel to the monocular camera is also higher, resulting in a more accurate depth image; that is, the depth values ​​of each pixel in the generated depth image are highly accurate.

[0101] Then, a first region is determined, consisting of pixels with depth values ​​greater than a distance threshold in the depth image. Since this region is far from the humanoid robot, it will not move to such a distant region. Therefore, the first region is considered an impassable region determined by the depth image. Furthermore, to ensure that the final traversable region corresponds to the actual spatial region, the first region in the depth image is mapped onto the original image to obtain a second region corresponding to the first region. Since the humanoid robot moves vertically, the rectangular region defined by the width of the second region and the height of the original image is determined as the impassable region of the original image. Based on this impassable region, the traversable region of the humanoid robot is determined in the original image. Finally, the travel direction angle of the humanoid robot is determined, and the robot is controlled to move within the traversable region according to this angle.

[0102] As can be seen, the technical solution of this invention can accurately generate the passable area and travel direction angle of the humanoid robot in real time, and control the humanoid robot to travel within the passable area in real time according to the travel direction angle, thereby achieving the purpose of real-time navigation. Furthermore, the humanoid robot can still work stably in complex environments and scenarios such as complex lighting conditions, enhancing its robustness.

[0103] After describing the overall technical solution of this invention, the following section will introduce a method for generating a passable area for a humanoid robot, as provided in an embodiment of the present invention. Figure 3 As shown in the figure, the present invention provides a method for generating a passable area for a humanoid robot, which is applied to, for example... Figure 1 The humanoid robot shown includes a monocular camera and is equipped with a humanoid robot traversable area generation system. The method for generating the humanoid robot traversable area may include the following steps:

[0104] S310: When the humanoid robot is detected to be powered on and starting to move, the original image of the area to be traversed by the humanoid robot is acquired.

[0105] The original images were acquired in real time using a monocular camera. Furthermore, the original images were taken by the monocular camera of the area the humanoid robot was about to pass through, meaning the camera was positioned in the direction the robot was moving.

[0106] Specifically, upon detecting that the humanoid robot has been powered on and begun to move, the monocular camera is activated, and its communication is checked. If the monocular camera communication is normal, it means that the monocular camera is working properly and can acquire images in real time. At this point, the next step S320 is executed. If the monocular camera communication is abnormal, it means that the monocular camera is not working properly and cannot acquire images. In this case, the process continues to check whether the monocular camera communication is normal until it is determined that the monocular camera communication is normal.

[0107] Assuming the monocular camera is communicating normally, the image of the area to be traversed that the monocular camera can capture in real time can be referred to as the raw image for clarity.

[0108] S320 performs defocus restoration processing on the original image to obtain a defocus restored image, and generates a depth image based on the distance from each pixel in the defocus restored image to the monocular camera.

[0109] Since the original image directly captured by a monocular camera may be a defocused image with low clarity, in order to accurately generate the passable area of ​​the humanoid robot in subsequent steps, a defocus restoration algorithm can be called to perform defocus restoration processing on the original image, that is, to restore the clarity of the original image and obtain the defocused image.

[0110] After obtaining the defocused image, a depth estimation algorithm can be used to calculate the depth of the image, obtaining the distance from each pixel in the defocused image to the monocular camera, and forming a depth image with that distance as the depth. Because the defocused image has high clarity, the accuracy of the distances from each pixel to the monocular camera obtained by performing depth calculations on the defocused image is high, resulting in a highly accurate depth image. This helps in accurately determining the passable area of ​​the humanoid robot in subsequent steps.

[0111] In one implementation, performing defocus restoration processing on the original image to obtain a defocus-restored image may include the following steps:

[0112] The original image is input into a pre-trained defocused image restoration model to restore the original image from defocus and output the defocused image.

[0113] The defocused image restoration model is trained using pre-acquired defocused images and corresponding sharp images.

[0114] It should be noted that the specific implementation method of "performing defocus restoration processing on the original image to obtain a defocus-restored image" is as follows: Figure 2 The embodiments shown have already been described, and will not be repeated here.

[0115] In one implementation, generating a depth image based on the distance from each pixel in the defocused image to the monocular camera may include the following steps:

[0116] The image after defocus restoration is input into a pre-trained image depth prediction model so that the image depth prediction model can extract the image features of the image after defocus restoration, and based on the relationship between image features and depth distance, obtain the distance from each pixel in the image after defocus restoration to the monocular camera, and output a depth image with the distance as the depth.

[0117] The image depth prediction model is trained using clear images from an image library.

[0118] It should be noted that the specific implementation method of "performing defocus restoration processing on the original image to obtain a defocus-restored image" is as follows: Figure 2 The embodiments shown have already been described, and will not be repeated here.

[0119] S330, determine the first region composed of pixels in the depth image whose depth value is greater than the distance threshold, and map the first region onto the original image to obtain the second region.

[0120] The distance threshold is determined based on the humanoid robot's current travel speed and a preset duration.

[0121] Since areas with a distance greater than a distance threshold are far from the humanoid robot's current position, the robot will not move to these distant areas. Therefore, to improve the robot's response speed in determining traversable areas, a first region with pixel depth values ​​greater than the distance threshold can be identified in the depth image. This first region is the traversable region determined by the depth image. In practical applications, this traversable region is a rectangular area. The width of this rectangular region is the width of the area formed by pixels with depth values ​​greater than the distance threshold, and the height of this rectangular region can be the height of the depth image. There can be one or more traversable regions in the depth image.

[0122] The distance threshold is determined by multiplying the humanoid robot's current travel speed by a preset duration. For example, if the preset duration is 5 seconds, then the distance threshold is the current travel speed multiplied by 5 seconds.

[0123] To ensure that the finally determined passable areas correspond to the actual spatial areas, a first region is identified in the depth image. After identifying the impassable areas in the depth image, it is necessary to map the first region from the depth image onto the original image. Figure 2 As can be seen from the description of the illustrated embodiment, there is a one-to-one mapping relationship between the pixels of the depth image and the pixels of the original image. Therefore, after obtaining the first region in the depth image, the first region can be mapped to the original image according to the mapping relationship between the pixels of the depth image and the pixels of the original image to obtain the second region corresponding to the first region. This helps to accurately determine the impassable region in the original image in subsequent steps.

[0124] As one implementation of this disclosure, S330, determining the first region composed of pixels in the depth image whose depth values ​​are greater than a distance threshold, may include the following steps a1 to a3:

[0125] Step a1: Obtain the current passage speed and preset duration of the humanoid robot, and determine the product of the current passage speed and preset duration as the distance threshold.

[0126] Step a2: Locate target pixels in the depth image whose distance is greater than the distance threshold, and determine the width of the region formed by the target pixels.

[0127] Step a3: Based on the width of the region formed by the target pixels and the height of the depth image, determine the first region formed by pixels in the depth image whose depth value is greater than the distance threshold.

[0128] Specifically, the distance threshold is determined by multiplying the humanoid robot's current travel speed by a preset time. Assuming the preset time is 5 seconds, the distance threshold is the current travel speed multiplied by 5 seconds. Since areas with a distance greater than the distance threshold are far from the humanoid robot's current position, the robot will not move to these distant areas. Therefore, to improve the robot's response speed in determining passable areas, target pixels with depth values ​​greater than the distance threshold can be found in the depth image. The rectangular area defined by the width of the region formed by these target pixels and the height of the depth image is then defined as the first region, which is the impassable region determined in the depth image.

[0129] S340, Identify the impassable areas in the original image.

[0130] The impassable area is a rectangular region determined based on the width of the second region and the height of the original image.

[0131] Since the first region is a non-passable area in the depth image, and the second region is obtained by mapping the first region, the width of the second region is the width of the non-passable area in the original image. Furthermore, since the humanoid robot moves vertically, the rectangular area determined by the width of the second region and the height of the original image is the non-passable area in the original image. In practical applications, this non-passable area can be one or more.

[0132] S350, based on the impassable areas in the original image, determines the traversable areas of the humanoid robot in the original image.

[0133] Specifically, after identifying the impassable areas in the original image, the entire image can be subtracted from the impassable areas to obtain one or more rectangular regions. Passable regions can then be determined from these rectangular regions. In practical applications, some rectangular regions may be too narrow for the humanoid robot to pass through. Therefore, it is necessary to determine whether any rectangular regions exist with a width greater than a specified threshold.

[0134] The width threshold is determined by the width of the humanoid robot and a preset multiple. For example, the preset multiple could be 2, in which case the width threshold is the width of the humanoid robot multiplied by 2. Of course, in practical applications, the specific value of the preset multiple can be determined according to the actual situation, as long as it ensures that the humanoid robot can safely pass through the passable area.

[0135] In one implementation, determining the passable area of ​​the humanoid robot in the original image based on the impassable area can include the following steps, namely steps b1 to b4:

[0136] Step b1: Calculate the width of each rectangular region in the original image, excluding the impassable regions.

[0137] Step b2: Obtain the width of the humanoid robot and determine the width threshold by multiplying the width of the humanoid robot by a preset multiple.

[0138] Step b3: Compare the width of each rectangular region obtained in step b1 with the width threshold.

[0139] Step b4: If there is a target rectangular region with a width greater than the width threshold, the rectangular region with the largest width in the target rectangular region is determined as the passable area of ​​the humanoid robot.

[0140] Specifically, the width of each rectangular region in the original image, excluding the impassable areas, can be calculated. The width of each rectangular region can be compared with a width threshold. If there is a target rectangular region with a width greater than the width threshold, it means that the humanoid robot can safely and smoothly pass through the target rectangular region. Therefore, the target rectangular region can be determined as the passable area of ​​the humanoid robot.

[0141] In practical applications, there can be one or more target rectangular areas. If there is only one target rectangular area, it can be directly defined as the passable area for the humanoid robot. If there are multiple target rectangular areas, the widest rectangular area can be defined as the passable area for the humanoid robot, so that the humanoid robot can move within a spacious passable area in subsequent steps.

[0142] S360 determines the travel direction angle of the humanoid robot and controls the humanoid robot to travel within the passable area based on the travel direction angle.

[0143] Specifically, after determining the traversable area of ​​the humanoid robot, it is also necessary to determine the robot's direction of travel. This allows the robot to be controlled to move within the traversable area according to the specified direction and angle. Therefore, real-time navigation can be achieved by accurately determining the traversable area and direction of travel of the humanoid robot in real time.

[0144] To ensure clarity, the specific implementation of S360 will be described in detail in the following embodiments.

[0145] The technical solution provided in this invention uses a humanoid robot's monocular camera to acquire original images of the area to be traversed in real time. Since the original image may be a low-resolution, out-of-focus image, defocus restoration processing is performed to obtain a defocus-restored image. A depth image is then generated based on the distance from each pixel in the defocus-restored image to the monocular camera. Because the defocus-restored image has higher resolution, the accuracy of the distances from each pixel to the monocular camera is also higher, resulting in a more accurate depth image; that is, the depth values ​​of each pixel in the generated depth image are highly accurate.

[0146] Then, a first region is determined, consisting of pixels with depth values ​​greater than a distance threshold in the depth image. Since this region is far from the humanoid robot, it will not move to such a distant region. Therefore, the first region is considered an impassable region determined by the depth image. Furthermore, to ensure that the final traversable region corresponds to the actual spatial region, the first region in the depth image is mapped onto the original image to obtain a second region corresponding to the first region. Since the humanoid robot moves vertically, the rectangular region defined by the width of the second region and the height of the original image is determined as the impassable region of the original image. Based on this impassable region, the traversable region of the humanoid robot is determined in the original image. Finally, the travel direction angle of the humanoid robot is determined, and the robot is controlled to move within the traversable region according to this angle.

[0147] As can be seen, the technical solution of this invention can accurately generate the passable area and travel direction angle of the humanoid robot in real time, and control the humanoid robot to travel within the passable area in real time according to the travel direction angle, thereby achieving the purpose of real-time navigation. Furthermore, the humanoid robot can still work stably in complex environments and scenarios such as complex lighting conditions, enhancing its robustness.

[0148] exist Figure 3 Based on the illustrated embodiments, as Figure 4 As shown, in one embodiment, S360, determining the travel direction angle of the humanoid robot and controlling the humanoid robot to travel within the passable area based on the travel direction angle may include the following steps:

[0149] S361, determine the coordinates of the first center pixel of the passable area, and determine the coordinates of the second center pixel of the original image.

[0150] Specifically, in order to calculate the travel direction angle of the humanoid robot, after determining the passable area, it is necessary to determine the coordinates of the center pixel of the passable area. For clarity of the scheme description, this center pixel is referred to as the first center pixel, and its coordinates are denoted as (x, y). The coordinates of the center pixel of the original image are also determined. For clarity of the scheme description, this center pixel is referred to as the second center pixel, and its coordinates are denoted as (x0, y0).

[0151] S362, determine the depth value of the first center pixel.

[0152] The depth value of the first center pixel is the depth value of the center pixel of the depth image.

[0153] Since there is a one-to-one mapping between the pixels of the original image and the pixels of the depth image, and the first center pixel is mapped to the center pixel of the depth image, the depth value of the center pixel of the depth image can be determined as the depth value of the first center pixel, which can be denoted as d.

[0154] S363 calculates the travel direction angle of the humanoid robot based on the coordinates of the first center pixel, the coordinates of the second center pixel, and the depth value of the first center pixel.

[0155] After obtaining the coordinates (x, y) of the first center pixel, the coordinates (x0, y0) of the second center pixel, and the depth value d of the first center pixel, arcsin(x-x0) / d can be used as the travel direction angle of the humanoid robot.

[0156] S364, according to the humanoid robot's direction of travel and current travel speed, control the humanoid robot to travel in the passable area, and return to execute the step of acquiring the original image of the humanoid robot's area to be traveled.

[0157] Once the travel direction angle of the humanoid robot is determined, its travel direction can be determined. Furthermore, the humanoid robot can maintain its current travel speed. In other words, the humanoid robot is controlled to travel within the passable area according to its travel direction angle and current travel speed, and then the process returns to execute step S310 to obtain the original image of the area to be traversed by the humanoid robot.

[0158] As can be seen, the technical solution of this invention can generate the passable area of ​​the humanoid robot in real time and accurately, and can also determine the passable angle and passable speed of the humanoid robot in real time and accurately, thus achieving the purpose of real-time navigation.

[0159] exist Figure 3Based on the illustrated embodiment, in one implementation, the method for generating a traversable area for the humanoid robot may further include the following steps:

[0160] If there is no passable area for the humanoid robot in the original image, control the humanoid robot to rotate in place by a preset angle and control the humanoid robot's passing speed to zero.

[0161] Return to the step of acquiring the original image of the area to be traversed by the humanoid robot.

[0162] Specifically, if the original image contains only impassable areas, meaning the number of remaining passable areas is 0, it indicates that the humanoid robot has no passable areas. In this case, the humanoid robot can be controlled to rotate in place by a preset angle. This preset angle can be set according to the actual situation and is not specifically limited here. For example, the preset angle can be +45°, indicating that the humanoid robot can turn 45° clockwise in place; the preset angle can be -45°, indicating that the humanoid robot can turn 45° counterclockwise in place. After controlling the humanoid robot to rotate in place by the preset angle, return to step S220, that is, regenerate the passable areas of the humanoid robot.

[0163] As can be seen, with the technical solution provided in this embodiment, if there is currently no passable area, the humanoid robot can be controlled to rotate in place, and the passable area of ​​the humanoid robot can be re-determined, that is, the route of the humanoid robot can be re-planned.

[0164] Secondly, embodiments of the present invention provide a humanoid robot traversable area generation device, applicable to humanoid robots including a monocular camera, such as... Figure 5 As shown, the device includes:

[0165] The original image acquisition module 510 is used to acquire the original image of the area to be traversed by the humanoid robot when the robot is detected to be powered on and starting to move. The original image is acquired in real time by the monocular camera.

[0166] The depth image generation module 520 is used to perform defocus restoration processing on the original image to obtain a defocus restored image, and generate a depth image based on the distance from each pixel in the defocus restored image to the monocular camera.

[0167] The region determination module 530 is used to determine a first region composed of pixels in the depth image whose depth values ​​are greater than a distance threshold, and to map the first region onto the original image to obtain a second region; the distance threshold is determined based on the current travel speed of the humanoid robot and a preset time.

[0168] The impassable area determination module 540 is used to determine the impassable area of ​​the original image, wherein the impassable area is a rectangular area determined based on the width of the second area and the height of the original image;

[0169] The passable area determination module 550 is used to determine the passable area of ​​the humanoid robot in the original image based on the impassable areas in the original image.

[0170] The humanoid robot control module 560 is used to determine the travel direction angle of the humanoid robot and control the humanoid robot to travel in the passable area based on the travel direction angle.

[0171] Optionally, the region determination module is specifically used for:

[0172] The current travel speed of the humanoid robot is obtained, and the product of the current travel speed and the preset time is determined as the distance threshold;

[0173] In the depth image, locate target pixels whose distance is greater than the distance threshold, and determine the width of the region formed by the target pixels;

[0174] Based on the width of the region formed by the target pixels and the height of the depth image, a first region is determined, which is composed of pixels in the depth image whose depth value is greater than a distance threshold.

[0175] Optionally, the passable area determination module is specifically used for:

[0176] Calculate the width of each rectangular region in the original image, excluding the impassable region;

[0177] The width of the humanoid robot is obtained, and the product of the width of the humanoid robot and a preset multiple is determined as the width threshold.

[0178] The width of each rectangular region is compared with the width threshold.

[0179] If there is a target rectangular region with a width greater than the width threshold, the rectangular region with the largest width in the target rectangular region is determined as the passable area of ​​the humanoid robot.

[0180] Optionally, the humanoid robot control module is specifically used for:

[0181] Determine the coordinates of the first center pixel of the passable area, and determine the coordinates of the second center pixel of the original image;

[0182] Determine the depth value of the first center pixel, wherein the depth value of the first center pixel is the depth value of the center pixel of the depth image;

[0183] Based on the coordinates of the first center pixel, the coordinates of the second center pixel, and the depth value of the first center pixel, the travel direction angle of the humanoid robot is calculated;

[0184] According to the humanoid robot's travel direction angle and current travel speed, control the humanoid robot to travel in the passable area, and return to execute the step of acquiring the original image of the humanoid robot's area to be traveled.

[0185] Optionally, the humanoid robot control module is further used for:

[0186] If there is no passable area for the humanoid robot in the original image, control the humanoid robot to rotate in place by a preset angle and control the humanoid robot's passing speed to zero;

[0187] Return to the step of obtaining the original image of the area to be traversed by the humanoid robot.

[0188] Optionally, the depth image generation module is specifically used for:

[0189] The original image is input into a pre-trained defocused image restoration model to perform defocus restoration processing on the original image and output the defocused image restored.

[0190] The defocused image restoration model is obtained by training with pre-acquired defocused images and corresponding sharp images.

[0191] Optionally, the depth image generation module is specifically used for:

[0192] The image after defocus restoration is input into a pre-trained image depth prediction model so that the image depth prediction model extracts the image features of the image after defocus restoration. Based on the relationship between image features and depth distance, the distance from each pixel in the image after defocus restoration to the monocular camera is obtained, and a depth image with the distance as the depth is output.

[0193] The image depth prediction model is trained using clear images from an image library.

[0194] Thirdly, embodiments of the present invention provide a humanoid robot 600, such as... Figure 6 As shown, it includes:

[0195] The monocular camera 601 is used to acquire raw images of the area to be traversed by the humanoid robot in real time;

[0196] At least one processor 602 is configured to acquire the raw image from the monocular camera upon detecting that the humanoid robot has been powered on and begun to move;

[0197] Memory 603 for storing the at least one processor-executable instruction;

[0198] The at least one processor is configured to execute the instructions to implement the traversable area generation method for the humanoid robot described in the above embodiments.

[0199] Fourthly, embodiments of the present invention provide a computer-readable storage medium that, when the instructions in the computer-readable storage medium are executed by a processor of an electronic device, enables the electronic device to perform the humanoid robot traversable area generation method described in the above embodiments.

[0200] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention without departing from the principles and spirit of the present invention.

Claims

1. A method for generating a passable area for a humanoid robot, characterized in that, Applied to a humanoid robot including a monocular camera, the method includes: When the humanoid robot is detected to be powered on and starting to move, the original image of the area to be traversed by the humanoid robot is acquired, and the original image is acquired in real time by the monocular camera; The original image is defocused and restored to obtain a defocused image, and a depth image is generated based on the distance from each pixel in the defocused image to the monocular camera. A first region is determined, consisting of pixels in the depth image whose depth values ​​are greater than a distance threshold, and the first region is mapped onto the original image to obtain a second region; the distance threshold is determined based on the current travel speed of the humanoid robot and a preset time. The impassable region of the original image is determined as a rectangular region based on the width of the second region and the height of the original image. Based on the impassable areas in the original image, the traversable areas of the humanoid robot are determined in the original image; Determine the travel direction angle of the humanoid robot, and control the humanoid robot to travel in the passable area based on the travel direction angle; Determining the first region composed of pixels in the depth image whose depth values ​​are greater than a distance threshold includes: The current passage speed and preset duration of the humanoid robot are obtained, and the product of the current passage speed and the preset duration is determined as the distance threshold. In the depth image, locate target pixels whose distance is greater than the distance threshold, and determine the width of the region formed by the target pixels; Based on the width of the region formed by the target pixels and the height of the depth image, a first region is determined, which is composed of pixels in the depth image whose depth value is greater than a distance threshold. The determination of the passable area of ​​the humanoid robot in the original image, based on the impassable area of ​​the original image, includes: Calculate the width of each rectangular region in the original image, excluding the impassable region; The width of the humanoid robot is obtained, and the product of the width of the humanoid robot and a preset multiple is determined as the width threshold. The width of each rectangular region is compared with the width threshold. If there is a target rectangular region with a width greater than the width threshold, the rectangular region with the largest width in the target rectangular region is determined as the passable area of ​​the humanoid robot.

2. The method for generating a passable area for a humanoid robot according to claim 1, characterized in that, Determining the travel direction angle of the humanoid robot, and controlling the humanoid robot to travel within the passable area based on the travel direction angle, includes: Determine the coordinates of the first center pixel of the passable area, and determine the coordinates of the second center pixel of the original image; Determine the depth value of the first center pixel, wherein the depth value of the first center pixel is the depth value of the center pixel of the depth image; Based on the coordinates of the first center pixel, the coordinates of the second center pixel, and the depth value of the first center pixel, the travel direction angle of the humanoid robot is calculated; According to the humanoid robot's travel direction angle and current travel speed, control the humanoid robot to travel in the passable area, and return to execute the step of acquiring the original image of the humanoid robot's area to be traveled.

3. The method for generating a passable area for a humanoid robot according to claim 1, characterized in that, The method further includes: If there is no passable area for the humanoid robot in the original image, control the humanoid robot to rotate in place by a preset angle and control the humanoid robot's passing speed to zero; Return to the step of obtaining the original image of the area to be traversed by the humanoid robot.

4. The method for generating a passable area for a humanoid robot according to claim 1, characterized in that, The process of restoring focus on the original image to obtain a defocused image includes: The original image is input into a pre-trained defocused image restoration model to perform defocus restoration processing on the original image and output the defocused image restored. The defocused image restoration model is obtained by training with pre-acquired defocused images and corresponding sharp images.

5. The method for generating a passable area for a humanoid robot according to claim 1, characterized in that, The process of generating a depth image based on the distance from each pixel in the defocused image to the monocular camera includes: The image after defocus restoration is input into a pre-trained image depth prediction model so that the image depth prediction model extracts the image features of the image after defocus restoration. Based on the relationship between image features and depth distance, the distance from each pixel in the image after defocus restoration to the monocular camera is obtained, and a depth image with the distance as the depth is output. The image depth prediction model is trained using clear images from an image library.

6. A device for generating a passable area for a humanoid robot, characterized in that, The apparatus is used to perform the humanoid robot traversable area generation method according to any one of claims 1 to 5, the apparatus being applied to a humanoid robot including a monocular camera, the apparatus comprising: The original image acquisition module is used to acquire the original image of the area to be traversed by the humanoid robot when the robot is detected to be powered on and starting to move. The original image is acquired in real time by the monocular camera. The depth image generation module is used to perform defocus restoration processing on the original image to obtain a defocus restored image, and generate a depth image based on the distance from each pixel in the defocus restored image to the monocular camera. The region determination module is used to determine a first region composed of pixels in the depth image whose depth values ​​are greater than a distance threshold, and to map the first region onto the original image to obtain a second region; the distance threshold is determined based on the current travel speed of the humanoid robot and a preset time. An impassable region determination module is used to determine impassable regions of the original image, wherein the impassable region is a rectangular region determined based on the width of the second region and the height of the original image; A passable area determination module is used to determine the passable area of ​​the humanoid robot in the original image based on the impassable areas in the original image; The humanoid robot control module is used to determine the travel direction angle of the humanoid robot and, based on the travel direction angle, control the humanoid robot to travel in the passable area.

7. A humanoid robot, characterized in that, include: A monocular camera is used to capture raw images of the area to be traversed by the humanoid robot in real time. At least one processor is configured to acquire the raw image from the monocular camera upon detecting that the humanoid robot has been powered on and begun to move; Memory for storing the at least one processor-executable instruction; The at least one processor is configured to execute the instructions to implement the humanoid robot navigable area generation method as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, When the instructions in the computer-readable storage medium are executed by the processor of the electronic device, the electronic device is able to perform the humanoid robot navigable area generation method as described in any one of claims 1-5.

Images