53 results about "Object point" patented technology

The invention relates to a robot grabbing control method and device, computer equipment, a computer readable storage medium and a computer program product. The method comprises the following steps: acquiring an object image and a depth image collected for a capture area; generating three-dimensional point cloud information of the capture area based on the depth image; for each grabbed object, carrying out grabbed point prediction based on the object point cloud information and the object information of the grabbed object to obtain a plurality of grabbed points predicted for the grabbed object and respective grabbed point information of each grabbed point; for each grabbing point, carrying out information comparison on the grabbing point information of the grabbing point and other point cloud information, and taking the grabbing point of which the comparison result meets a comparison condition as a target grabbing point; and executing a grabbing operation according to the target grabbing point information of the target grabbing point. By adopting the method, the grabbing accuracy can be improved.

The application discloses a business risk measurement method, and belongs to the computer field. The method comprises the following steps: obtaining a text map picture and a satellite map picture of a target enterprise; obtaining a target text on the text map picture and a target object on the satellite map picture; matching the target text and the target object to determine whether the target text and the target object point to the same thing; if the target text and the target object point to the same thing, obtaining a first type to which the thing belongs; if the target text and the target object point to different things, obtaining a second type to which the target text belongs and a third type to which the target object belongs; and obtaining a business risk assessment result for the target enterprise according to at least one of the first type to which the thing belongs, the second type to which the target text belongs and the third type to which the target object belongs.

A target path changing system for an accessory device is provided, which can improve work efficiency of an operation to change a target path of a specific portion of the accessory device. An object point selection unit (15) is configured to allow an operator to select one of target points (72) as a change object point (73). A range setting unit (15) is configured to set a movable range (E) in which the change object point (73) is movable. An object point moving unit (15) allows the operator to move the change object point (73) within the movable range (E). A target path resetting unit (15) is configured to reset the target path (71a) between a change start point (74) and a change end point (75) so that the target path (71) passes through a post-change object point (73a).

To adjust the advantage of a combination of a plurality of objects selected by a player in the progress of a play according to a use state.SOLUTION: The information processing device includes a use frequency acquisition unit configured to acquire a use frequency of each object, the use frequency being determined based on the number of times an object usable in a play is selected by a player, a point setting unit configured to set a point of each object based on the use frequency, and a parameter value determination unit configured to determine at least one parameter value related to progress of a play of the player based on a point of one or more objects satisfying a predetermined condition among a plurality of objects selected by the player.SELECTED DRAWING: Figure 3

A point cloud data processing apparatus 11 includes a processor configured to acquire first form information that indicates a feature of a form of a first object, specify an object region of a second object that is identified from an image and that corresponds to the first form information, select second-object point cloud data, in point cloud data, that corresponds to the object region, on the basis of the object region, acquire second form information that indicates a feature of a form of the second object, on the basis of the second-object point cloud data, and compare the first form information with the second form information and perform determination as to whether the second object is the first object.

The application discloses a kind of dual-hand dexterous grasping data synthesis and grasping generation method, system and robot for multi-size object.Data synthesis method is: local grasping area is constructed on object surface, target area pair is determined and posture is initialized after geometric screening and mechanical stability evaluation;Decoupling optimization objective function including local force closure, contact distance and area consistency term is constructed, and left and right hand postures are decoupled and optimized;Feasible sample is screened and motion trajectory is generated by collision detection and inverse kinematics solution, and large-scale data set is constructed.Grasping generation method is: based on object point cloud cooperative prediction master-slave grasping perspective;Scale self-adaptive grasping anchor point is determined;Local features and global features are fused to construct conditional features;Generation model is used to generate dual-hand coordinated grasping posture in the form of pose increment relative to anchor point.The application significantly improves data generation efficiency and quality, and enhances the generalization ability of model to multi-size object and the executability of grasping.

The invention relates to an apparatus for generating shape data of an object and a method for generating shape data of an object. A computer device (1) is provided with: a first arithmetic unit (10) that outputs a numerical value obtained on the basis of coordinate values of object points indicating the positions of each part of the shape of an object; and a second calculation means (20) that, when a numerical value is input, outputs coordinate values of object points indicating the positions of the respective parts of the shape of the object and physical quantities at the respective object points such that coordinate values of learning object points indicating the shape of the learning object match the coordinate values of the object points output by the second calculation means. The second arithmetic unit learns the physical quantity output by the second arithmetic unit so as to satisfy an equation indicating the physical law that the physical quantity should satisfy, and the learned second arithmetic unit outputs the coordinate value of the object point and the physical quantity following the physical law on the basis of the input numerical value.

This disclosure provides a point cloud processing method, apparatus, computing device, and storage medium. The point cloud processing method includes: acquiring object states in an initial point cloud frame; extracting object point clouds from the initial point cloud frame based on the object states; calculating a shape code of the initial point cloud frame based on a shape model and the object point clouds in the initial point cloud frame, wherein the shape code is used to characterize the surface shape of the object; and calculating the object states and / or shape codes of corresponding objects in other point cloud frames of the point cloud sequence based on the shape code and object states of the initial point cloud frame. Embodiments of this disclosure combine object tracking and shape reconstruction based on a shape model, improving the accuracy of object tracking and the fidelity of shape reconstruction.

The present application belongs to the field of visual tracking technology, and discloses a kind of processing existing partial occlusion flexible linear object visual tracking method, based on RGB-D camera obtains color image and depth image data, from first frame image, by mask segmentation, skeletonization processing operation generates initial state point, in each subsequent frame, object point cloud is generated by mask in conjunction with depth image;Then, the density of its surrounding current point cloud is counted, and the state point is marked as "visible" or "occlusion";By establishing the analytical model based on the spatiotemporal continuity of flexible linear object, the closed form solution of the occlusion position is derived, and the iterative optimization process of the occlusion position is eliminated, thereby reducing the computational complexity and achieving efficient real-time tracking;Meanwhile, using the inherent spatiotemporal geometric evolution law of flexible linear object, a position estimation model that can be directly analytically solved is constructed, avoiding the stability risk brought by iterative optimization, and significantly improving the reliability and tracking stability of the occlusion position estimation.

Disclosed herein are system, method, and computer program product embodiments for localizing three-dimensional objects relative to a vehicle. The system includes: at least one sensor for generating two-dimensional (2D) data and a three-dimensional (3D) point cloud of an environment external to a vehicle. The 3D point cloud includes object points associated with a stationary traffic control object. The localization system also includes a memory and at least one processor coupled to the memory. The processor is programmed to: select a bounding box associated with the object from the memory based on the 2D data; arrange the bounding box proximate to the object points in the 3D point cloud; assign a weight to each point of the 3D point cloud based on a position of the point relative to the bounding box; filter the weighted points; and generate a 3D location of the object based on the filtered points.

The invention belongs to the technical field of visual tracking, and discloses a visual tracking method for processing a partially-shielded flexible linear object, which comprises the following steps of: acquiring color image and depth image data based on an RGB-D camera, generating an initial state point from a first frame of image through mask segmentation and skeletonization processing operation, and in each subsequent frame, generating an initial state point from a second frame of image through mask segmentation and skeletonization processing operation; generating an object point cloud by combining the mask with the depth image; then, counting the density of current point clouds around the state point, and marking the state point as visible or shielded; an analytical model based on space-time continuity of a flexible linear object is established, a closed form solution of a shielding position is deduced, and an iterative optimization process of a shielding state point position is eliminated, so that the calculation complexity is reduced, and efficient real-time tracking is realized; and meanwhile, the inherent space-time geometric evolution law of the flexible linear object is utilized to construct a position estimation model which can be directly analyzed and solved, so that the stability risk caused by iterative optimization is avoided, and the reliability and tracking stability of shielding position estimation are remarkably improved.

In some cases, a calibration method, calibration object, device, readable storage medium, and program product are provided. The method includes: acquiring first image data, which is obtained by a camera capturing an object in three-dimensional space; the object includes multiple object points, and the first image data includes first coordinates of the multiple object points in a first coordinate system; reprojecting second coordinates of the object points onto the first coordinate system based on a distorted camera model to obtain third coordinates; wherein, during the process of reprojecting the second coordinates onto the first coordinate system, the second coordinates are mapped onto a preset plane to obtain fourth coordinates, and the fourth coordinates are distorted using a first distortion parameter; the first distortion parameter is obtained by interpolation of second distortion parameters of multiple control points, the multiple control points being located in different regions of the preset plane, and the second coordinates being the coordinates of the object points in the second coordinate system; and determining camera parameters based on the third coordinates and the first coordinates, the parameters including the second distortion parameter.

The invention provides a motion posture determination method and device, electronic equipment and a storage medium, and relates to the technical field of artificial intelligence, in particular to the technical field of computer vision. Dividing the first to-be-processed point cloud to obtain a first object point cloud set for representing a target object and a field point cloud set for representing a sports field; in response to the fact that the distance between the point with the minimum target dimension value in the first object point cloud set and the target plane is smaller than a first preset threshold value, the collection moment of the first to-be-processed point cloud is determined as the motion starting moment of the target object; according to the angle of the fitting line segment of the first object point cloud set relative to the target plane, determining the motion posture of the target object at the motion starting moment; the fitting line segment is obtained by performing linear fitting on the first object point cloud set.

The application relates to an image processing method and device and a computer device. The method comprises the following steps: obtaining a historical object image set and a current object image set corresponding to a target object; performing object point matching based on the historical object image set and the current object image set to obtain historical object matching points and current object matching points; when the historical object matching points and the current feature matching points are the same object points, performing spatial pose transformation parameter calculation on the historical object matching points to obtain historical spatial pose transformation parameters, and performing three-dimensional space matching point calculation to obtain historical feature point three-dimensional coordinates; performing spatial pose transformation parameter calculation on the current feature matching points to obtain current spatial pose transformation parameters, and performing three-dimensional space matching point calculation to obtain current feature point three-dimensional coordinates; and calculating a difference value corresponding to the historical feature point three-dimensional coordinates and the current feature point three-dimensional coordinates to obtain three-dimensional offset information corresponding to the target object. The method can improve the accuracy of deformation monitoring.

The embodiment of the application discloses a kind of grasping pose estimation method and related equipment, it is characterized in that, the method includes: based on the M sample points in the object point cloud of target object, determine the M sample point clouds of the object point cloud, each sample point cloud in the M sample point clouds is the point cloud of the object point cloud in corresponding target space, the corresponding target space is the spherical space with the M sample points in corresponding sample point as ball center and with preset length as radius;Wherein, the number of points included in each sample point cloud is greater than preset value;Based on the M sample point clouds and grasping end model parameter, M grasping poses are calculated, and the grasping end model parameter includes the width and depth of grasping end model;Select one grasping pose from the M grasping poses as the target grasping pose for grasping the target object.Using the embodiment of the application can improve the robot grasping efficiency, accuracy in complex scene.

The embodiment of the specification provides a feature processing method, wherein the feature processing method comprises: acquiring an initial feature sequence, and selecting a feature to be masked in the initial feature sequence; performing mask processing on the feature to be masked, generating an intermediate feature sequence according to a mask processing result, and predicting a predicted feature corresponding to the feature to be masked according to the intermediate feature sequence; calculating a predicted feature score corresponding to the predicted feature, and updating the initial feature sequence into a target feature sequence according to the predicted feature score and feature scores respectively corresponding to features in the intermediate feature sequence; and inputting the target feature sequence into an object click prediction model for processing to obtain object click prediction information.

A method, system and storage medium for object detection are presented. The method includes; creating object representation data based on outputs from a neural network and augmented reality (AR); constructing a semantic map comprising object superpoints; determining a set of object superpoints in the semantic map whose locations are within a certain distance of a new incoming object point; updating the semantic map in response to the new incoming object point; and modifying a probability of a class to which the new incoming object point belongs as identified by the neural network based on the updated semantic map. The semantic map is composed of object superpoints having a list of scores corresponding to detected labels, a list of viewing directions, and a list of sizes. Based on the semantic map, the accuracy of object detection is improved by modifying the probability of an object label or class from the neural network.

The camera generates a learning image by photographing an object point, and acquires a plurality of learning parameters related to the position and posture of the camera when the learning image is photographed (S1). A model is generated which takes the plurality of learning parameters as input and outputs a first displacement amount of an object block corresponding to the object point displayed in the learning image with respect to a reference position in the learning image (S2). The camera generates a measurement image by photographing the object point, and acquires a plurality of measurement parameters related to the position and posture of the camera when the measurement image is photographed (S3). A second displacement amount of the object block corresponding to the object point displayed in the measurement image with respect to the reference position in the measurement image is output by inputting the plurality of measurement parameters into the above model (S4). A displacement obtained by subtracting the second displacement amount from the displacement of the object block in the measurement image is output (S5, S6).

The embodiment of the application provides a kind of predict object availability and the method of robot action and related equipment, belong to the cross technical field of robot control and artificial intelligence.The method comprises: receiving input containing task instruction;Based on the task instruction, the scene is parsed by visual language model, and the object point cloud of at least one object related to task is acquired;The object point cloud and the task instruction are input into the diffusion model trained in advance;Denoising inference is carried out by the diffusion model, and at least one 6 degree of freedom availability pose indicating that robot interacts with the object is output, and the availability pose includes three-dimensional translation and three-dimensional rotation information.The application overcomes the problems of availability estimation method, such as lack of orientation information, dependence on additional modules, etc., can provide accurate, robust and directly usable for robot control 6D pose for embodied tasks, and improves the operation ability and automation level of robot in complex scene.

The embodiment of the present application provides a kind of modeling method, system, equipment and storage medium for extracting foreground object, belong to three-dimensional reconstruction technical field.Modeling method for extracting foreground object includes: obtaining the image point cloud data of target object and camera pose information;The foreground and background in image point cloud data are separated, and the foreground object point cloud is obtained;According to the foreground object point cloud extracted, the minimum bounding box of foreground is determined;Based on the above data, foreground point data and background point data are determined;Based on foreground point data and background point data, the preset reconstruction network is used to determine foreground object reconstruction data and background reconstruction data;According to foreground object reconstruction data and background reconstruction data, generate target object model and display.The embodiment of the present application calibrates foreground and background using minimum bounding box, reduces the cost of manual annotation, also uses preset reconstruction network to predict model data, improves the accuracy and accuracy of model reconstruction.

The invention discloses a two-hand combined grabbing posture synthesis method and system based on a nerve grabbing distance field, and belongs to the technical field of robots. The method comprises the following steps: acquiring object point cloud data and constructing training data based on a single-hand grabbed data set; extracting shape embedding features of the point cloud by training an object feature extraction network; further training a neural grabbing distance field network, and outputting a grabbing distance vector including a single-hand grabbing distance, a double-hand penetration distance and a double-hand balance distance by taking object shape features and double-hand query postures as input; and finally, the distance vector is minimized through gradient optimization, iterative updating of the two-hand query posture is achieved, and the two-hand optimal grabbing posture is generated. According to the method, two-hand grabbing synthesis can be achieved only by grabbing data with a single hand, data dependence and cost are remarkably reduced, grabbing stability and collision-free performance are guaranteed by fusing various distance constraints, and the method can be applied to robot task scenes needing two-hand coordinated operation such as family service and industrial carrying.

The present application provides a point cloud data processing device, a point cloud data processing method and a program, which can obtain the correctness of the determination result when determining the region of the object as the detection object in the image, and can effectively utilize the determination result. In the point cloud data processing device (11), the processor acquires first shape information representing the shape feature of the first object, determines the object region of the second object corresponding to the first shape information according to the image discrimination, selects the second object point cloud data corresponding to the object region in the point cloud data based on the object region, acquires second shape information representing the shape feature of the second object based on the second object point cloud data, compares the first shape information and the second shape information, and determines whether the second object is the first object.

The invention relates to an object volume determination method and device, computer equipment and a readable storage medium, and relates to the technical field of logistics. The method comprises the following steps: acquiring an image of an object; the image comprises a reference object area, and the reference object area is obtained by frame selection of an object in the image; determining all planes in the image, determining an intersecting plane intersecting with the reference object area in each plane based on the similarity between each plane and the reference object area, and determining a support plane of the object based on the position relationship between the intersecting plane and the reference object area; determining an object point cloud of the object according to the intersecting plane and the reference object area; and determining the volume of the object according to the plane point cloud of the supporting plane and the object point cloud. By adopting the method, the object volume determination efficiency can be improved.

The invention belongs to the technical field of intelligent control, and discloses a robot non-grabbing external smart operation method based on rigidity perception dual-stage reinforcement learning, and the method comprises the steps: obtaining an environment and object point cloud through a depth camera, and obtaining a serialized tactile point cloud in combination with tentative tactile contact to perceive the object rigidity; multi-source point cloud information is fused through a high-level agent, and a target pose and an interaction action type are decided; the bottom intelligent agent plans a specific contact point and a motion direction according to the motion type; and the robot updates the state after executing the action, and circulates until the target is achieved. According to the method, the object rigidity information is fused into the decision process, the self-adaptive operation from rigid objects to soft objects is supported, the strategy does not need to be pre-defined manually, and the method has good generalization ability and learning efficiency and is suitable for robot flexible operation in a complex real scene.

The present disclosure relates to the technical field of fringe projection three-dimensional reconstruction, and particularly relates to a point cloud enhancement method and system based on polarization structure perception and trend guided recovery; the method comprises: acquiring polarization images at multiple angles, generating an object region mask based on a polarization angle consistency mask, a polarization angle gradient mask and a linear polarization degree mask; obtaining an original phase image through four-step phase shifting and a complementary Gray code, and then obtaining an internal missing region of an object according to a hole distribution of the original phase image, adding a morphological structure index and a local variance feature to identify a real self-shadow region; for the identified real self-shadow region, a structure guided recovery model combining a first-order structure gradient guide term and a second-order structure gradient guide term is established to obtain a complete object point cloud; the background and invalid shadow in the object point cloud are separated through the object region mask to obtain a point cloud retaining only the object region.

The application discloses an object grabbing method and device, equipment and a storage medium, and relates to the technical field of object grabbing. The method comprises the following steps: determining a local extreme point from a point cloud of a target object according to a point cloud access vector; wherein the elements in the point cloud access vector correspond to the points in the point cloud; the element value of the element is used to represent whether the point corresponding to the element in the point cloud is accessed; determining a local point cloud according to the local extreme point; and determining the grabbing pose of the robot according to the local point cloud, so as to realize the grabbing of the target object. The technical scheme of the embodiment of the application determines the local extreme point, further determines the local point cloud, and determines the grabbing pose of the robot according to the local point cloud, thereby improving the efficiency of determining the local extreme point in the object point cloud and improving the efficiency of grabbing the object.

The application discloses a STEP file export method based on a string buffer mechanism, and belongs to the field of CAD model data processing.The method first constructs head information of the exported STEP file, and writes the information into a buffer array.Secondly, data information of the STEP file to be exported is constructed, and a pointer of a face* type is obtained.Then, a pointer of a loop* type and a pointer of a surface* type are obtained from an object pointed by the face* pointer.Finally, all data of the above-mentioned pointers are written into the buffer array and exported to a file in a specified path, so that the STEP file export is completed.The application realizes the STEP file export with the advantages of simplicity, high efficiency and low storage cost.

Method for aligning containers (2) in an alignment unit (1), comprising the steps: e) Providing a container to be aligned (2) at at least one measuring position (6) and imaging the container positioned in this way with at least one camera (3, 3') provided for container control on the alignment unit; f) Determining image coordinates (u, v) of at least one characteristic object point (17) of the depicted container; g) Providing at least one shape parameter (rb) characteristic of the surface shape of the container; h) Providing a coordinate transformation algorithm calculated using a method for calibrating the alignment unit (1) for the camera (3, 3'); and i) Transforming the world coordinates (xw, yw, zw) of the container into image coordinates (u, v) using the algorithm; and / or transforming the image coordinates (u, v) of the container into world coordinates (xw, yw, zw) using the algorithm, wherein the procedure for calibrating the alignment unit (1) comprises the following steps: a) Providing a calibration body (8) with several control points (9) at at least one calibration position (7) and imaging the control points positioned in this way with at least one camera (3, 3') provided for container control on the alignment unit; b) Determining world coordinates (xw, yw, zw) of the mapped control points; c) Determining image coordinates (u, v) of the depicted control points; d) Substituting the world coordinates and the image coordinates into a mathematical model of the camera image and calculating at least one algorithm for coordinate transformation from a camera coordinate system (15) of the camera (3, 3') into a world coordinate system (16) of the alignment unit (1), and wherein the value of the characteristic shape parameter (rb) is retrieved from a storage unit (4) to adapt the coordinate transformation to the type of container (2) to be aligned, wherein the retrieval is triggered automatically, in particular by a marker coupled to a container to be aligned.

The application provides a coordinate direction determination method and device of an object, and belongs to the technical field of robot vision. The coordinate direction determination method of the object comprises the following steps: obtaining target environment data, wherein the target environment data comprises a wall position and an object point cloud of a target object; determining a target bounding rectangle of the object point cloud, wherein the target bounding rectangle comprises intersecting first and second edges; determining a first distance value between the object point cloud and the target wall according to the wall position; and determining the coordinate direction of the target object according to the ratio of the length of the first edge to the length of the second edge and the first distance value.