45 results about "Loop closure" patented technology

The invention relates to a mapping system and method for flight inspection of indoor equipment, the system comprises a data acquisition end and an offline data processing end, the data acquisition end comprises an unmanned aerial vehicle and a sensing module, and the sensing module comprises a laser radar module, a binocular camera module, an inertial measurement module and an airborne recording module; the method comprises the following steps: synchronously acquiring multi-source data through layered flight; performing space-time alignment and preprocessing; self-adaptive sampling is carried out based on point cloud geometric prior guide image features, and laser-vision joint features are generated; fusing point cloud registration, joint features and inertial data, and carrying out joint estimation on adjacent frame pose increments; performing loopback detection based on the key frame to generate a closed-loop constraint; constructing and optimizing a pose map to eliminate cumulative drift; and finally fusing to generate a global point cloud map. According to the method, the space consistency and geometric accuracy of the map can be effectively improved, and a reliable data basis is provided for digital modeling and intelligent operation and maintenance of indoor equipment inspection.

This invention discloses a ground-air collaborative mapping method based on semantic features and ground plane multi-constraint fusion. The method includes: generating sub-map sets from local point cloud data collected by fire trucks and drones; and generating rotation-invariant semantic descriptors using a maximum height value rasterization strategy and rotation group enhancement. By calculating geometric and semantic distances, and combining isomorphic and heterogeneous pattern weighted fusion, candidate sub-map pairs with loop closures are selected and geometrically registered to generate loop closure constraints. Simultaneously, ground point clouds are extracted using the RANSAC algorithm, and the ground plane model is smoothly updated using a sliding window to construct height and pose alignment constraints. The loop closure constraints, ground plane constraints, and pose information are input into the Ceres optimization framework for joint optimization to generate a globally consistent overall map, effectively improving the accuracy and robustness of ground-air collaborative mapping, and is suitable for emergency scenarios such as complex fire rescue.

The present disclosure provides a method, an apparatus and a system for detecting a loop closure in an environment, the method executed by a processor, and comprising: identifying a text entity comprising a text content in the environment from an image frame associated with an image sensor; estimating a pose of the text entity in a current local coordinate frame associated with a depth sensor based on an estimated pose of the text entity in the image frame and an extrinsic parameter between the depth and image sensors; retrieving from a database using the text content a candidate pose of the text entity in a previous local coordinate frame; calculating a relative pose constraint relating to a transformation between the current and previous local coordinate frames based on the pose and the candidate pose of the text entity; and detecting the loop closure based on the relative pose constraint.

Examples described herein provide a method for trajectory estimation using an image sequence. The method includes receiving an image sequence of an environment from a camera moving relative to the environment. The method further includes extracting and matching features from images of the image sequence. The method further includes determining a relative orientation of the images of the image sequence. The method further includes determining orientation parameters of the camera using sequential image resection. The method further includes performing, using the orientation parameters, bundle adjustment to generate refined orientation parameters of the camera. The method further includes estimating a trajectory of the camera relative to the environment based at least in part on the refined orientation parameters by performing loop closure.

The present application relates to the technical field of computer vision, and discloses a loop closure detection method, an apparatus, a device, and a storage medium. The method comprises: determining a candidate key frame subset on the basis of an original image and a loop closure detection key frame database; when the candidate key frame subset is not empty, respectively determining a target image set for constructing a key frame and an original visual image; determining a matching point cloud pair by means of the target image set and the original visual image on the basis of a dense reconstruction thread, and calculating a similarity error of the matching point cloud pair; and performing loop closure detection on the basis of the similarity error. By means of said method, after a candidate key frame subset is determined, whether the candidate key frame subset is empty is determined; if so, it indicates that no loop closure is present, and if not, a dense reconstruction thread is created, and a valid matching point cloud pair is determined on the basis of the dense reconstruction thread; and then, loop closure detection is performed using an image pair corresponding to the similarity error, thereby effectively improving the accuracy of loop closure detection, improving SLAM tracking accuracy.

This invention discloses a dynamic environment SLAM localization method based on multi-source sensor data fusion, comprising the following steps: acquiring visual, inertial, and laser data and calibrating the buffer; performing HRNet segmentation and ORB-SLAM tracking on the visual data to obtain the initial pose; fusing the inertial and laser results to obtain the current pose; determining and recalculating the divergence of time window data to update the current pose; inserting keyframes and generating a semantic skeleton map; encoding and generating a semantic topology string and associating it with the keyframe index; retrieving loop closure candidate frames and performing geometric verification followed by map optimization to obtain the localization result and map result. This invention achieves highly robust localization and mapping in dynamic environments and improves the accuracy of loop closure recognition and map consistency.

The application discloses a kind of pose closed loop planning system and method based on PLC six visual quality inspection, it is related to industrial robot quality inspection technical field, including: visual perception module, for obtaining RGB-D image;Pose closed loop control module, for receiving RGB-D image and output 6D pose data;Motion planning module, for generating double-arm motion trajectory;Flip optimization module, for generating optimal flip sorting set;Sorting execution module, for completing sorting and discharging action according to quality inspection result and control instruction;Robot controller, for data interaction and issue control instruction.The application does not need to rely on artificial flip or multiple fixtures, reduce the failure rate caused by incoming material position deviation etc., through double-arm alternate operation and Monte Carlo optimization flip sorting, realize PLC device six automatic continuous detection, with the combination of coarse positioning, fine positioning of double RGB-D camera, combined with the output 6D pose data of deep learning pose estimation model, effectively avoid perspective distortion and miss detection, false detection.

The application relates to a loop detection method based on a point cloud semantic graph descriptor and position information, comprising the following steps: acquiring laser point cloud data and satellite positioning data of a current frame, performing semantic segmentation processing on the laser point cloud data, converting into a preset point cloud semantic graph descriptor form, screening a loop candidate frame from historical frame data through distance matching and node matching; converting the current frame and the loop candidate frame into a circular ring graph form, and attempting to obtain a target loop frame through similarity verification and geometric consistency verification. Compared with the prior art, the application solves the problems of difficult loop detection, large loop matching calculation amount and low loop matching accuracy in a mobile robot SLAM system, and improves the positioning accuracy and system robustness of the mobile robot.

This application provides a method, apparatus, electronic device, storage medium, and program product for robot map building. The method includes: constructing a map set based on environmental data collected during robot movement; constructing registration constraints for each sub-map based on the Euclidean symbolic distance field and surface voxel set of each sub-map; constructing odometry constraints for each sub-map based on the relative pose between any two adjacent sub-maps in the map set; performing loop closure detection on each sub-map based on loop closure information collected during robot movement, and constructing loop closure constraints for each sub-map; determining a reference pose map based on the loop closure information, and constructing global constraints for each sub-map based on the reference pose map; adjusting each sub-map based on the registration constraints, odometry constraints, loop closure constraints, and global constraints, and fusing the adjusted sub-maps to obtain a global map. This application improves the accuracy and consistency of map building with the environment.

This invention relates to a biomimetic brain-like synchronous localization and environmental perception method for underwater robots. The invention addresses the problems of poor visual odometry, insufficient robustness, and low accuracy in loop closure detection in existing underwater robot navigation systems. The robot first acquires environmental feature information through sonar sensors and its own motion information through navigation sensors. A local scene template is obtained by processing the sonar data using an acoustic-visual processing method. The sensor data undergoes pre-integration processing, and the processed data is used as input to a convolutional neural network to output motion displacement, thus forming the robot's perception of its own position. Finally, an empirical map integrates the above information, and loop closure detection updates the empirical map, reducing drift during robot movement and completing the construction of the empirical map. This invention belongs to the fields of bionics and motion navigation technology.

The application discloses a loop closure detection method based on block feature uniform weighting and distance sorting in outdoor complex environment, and belongs to the technical field of visual camera simultaneous localization and mapping and computer vision. The technical scheme of the application comprises the following steps: performing gray processing and Gaussian filtering pretreatment on the collected image pair, uniformly dividing the image into a plurality of sub-image blocks, extracting local feature descriptors from each sub-image block, calculating the distance between the feature vectors of the corresponding sub-image blocks based on cosine similarity, uniformly weighting the feature similarity of all sub-image blocks, sorting the calculated feature distances of all sub-image blocks from small to large, screening out the key matching block with the smallest distance by setting a threshold to eliminate the false matching block caused by illumination change noise, and comprehensively determining the global similarity of the feature distance of the key matching block to confirm the success of loop closure detection and output loop closure information to optimize the global pose of the visual camera.

The invention provides a positioning method, an electronic terminal and a computer readable storage medium, and the method comprises the steps: obtaining first point cloud data, and carrying out the compensation of the first point cloud data, and obtaining second point cloud data; matching the second point cloud data with a raster map to obtain a first matching result; and determining the pose of the target object based on the first matching result. According to the positioning method, the acquired first point cloud data is compensated, and the compensated second point cloud data is matched with the grid map in the loopback detection, so that the pose of the target object is determined, and the positioning precision of the target object in the loopback detection can be improved.

The invention relates to the field of computer vision, in particular to a pose closed-loop calibration method and system based on prism type screen splicing, and the method comprises the steps: collecting the relative pose data of a prism and each screen in real time, and synchronously obtaining the image data of a screen splicing region; calculating a picture distortion parameter of the image data according to the image data of the screen splicing area; if the picture distortion parameter exceeds a preset distortion threshold value, judging that the picture is abnormal; based on a judgment result, inputting the relative pose data and the picture distortion parameters into a preset error mapping model, determining a coupling error value of pose distortion through the error mapping model, and outputting a determined result; according to the coupling error value, analyzing a pose adjustment parameter of each screen; and based on the pose adjustment parameters of each screen, driving a pose adjustment mechanism installed on the display screen to compensate the relative position and angle of the prism and the screen.

This application relates to the field of computer vision technology and discloses a method and system for detecting pain expressions. The method includes: acquiring a facial video stream, segmenting muscle regions and extracting texture features after facial key point localization and inter-frame alignment; calculating motion energy based on these features, extracting enhanced micro-expression temporal segments and multi-scale spatiotemporal features, constructing a muscle dynamics model and completing state estimation; generating an adaptive candidate spatiotemporal window set through spatiotemporal attention weighted fusion; performing temporal modeling for dynamic feature fusion, and selecting the optimal window by fusing micro-expression features; and outputting the final pain level after double consistency verification and loop closure optimization. This application overcomes the limitations of static images, accurately captures facial dynamics and temporal changes, strengthens feature correlation, and significantly improves detection accuracy.

The invention belongs to the technical field of visual SLAM, and particularly relates to a visual SLAM method based on voxel grids and NeRF, and the method comprises the following steps: S1, carrying out the front-end tracking, and solving the posture transformation between images; s2, dynamic local mapping; s3, loopback detection is carried out; s4, designing a scene expression mode; s5, carrying out light ray sampling and volume rendering; the network architecture of front-end tracking in the step S1 comprises an optical flow estimation module and a neural network used for estimating the optical flow, and the neural network is composed of an encoder-decoder convolutional neural network and then used for adaptive camera attitude estimation; a camera pose coarse matching module initializes a pose estimation network using continuous images as input supervised training. According to the method, a micro chiral constraint attitude learning front-end method is combined with a neural radiation field, so that the generalization ability to an unfamiliar environment is relatively high, and the robustness is guaranteed; and the application effect in a complex environment is better.

This invention proposes a firefighting robot control method based on laser SLAM. The robot first uses Cartographer's SLAM algorithm to accurately extract image features from the disaster site, continuously building and refining the disaster site model based on these features. After receiving a firefighting mission, the robot first identifies flames based on information uploaded by its onboard depth camera. Once the flame source is identified, it actively plans a rescue route to reach the fire area as quickly as possible. Simultaneously, it uses its onboard binocular camera and LiDAR to avoid obstacles along the way, while infrared and distance sensors accurately identify the fire source, ensuring the robot quickly understands the situation on-site and can begin firefighting immediately upon arrival. This algorithm is based on an improved SLAM algorithm by Cartographer, incorporating a Lazy Decision algorithm in the loop closure detection part to optimize the Cartographer algorithm, effectively avoiding the consequences of loop closure errors and preventing incorrect loop closures.

The invention discloses a visual inertial positioning method, system and equipment for substation inspection and a storage medium. The method comprises the following steps: synchronously acquiring image data and inertial measurement unit data in a substation inspection process; extracting point features and line features from the current frame of image, and matching the point features and the line features with the previous frame of image to obtain a point and line feature corresponding relation; performing pre-integration on the data of the inertial measurement unit, and judging whether a new key frame is inserted or not by combining a visual observation result; constructing a joint optimization objective function in the sliding window, performing robust weighting on each residual item, and solving current state estimation through nonlinear optimization; and carrying out loopback detection by using a comprehensive matching result of the point features and the line features, and correcting the state quantity in the current sliding window to realize high-precision positioning. According to the method, the positioning robustness in a weak texture environment can be enhanced, abnormal values are inhibited, the anti-interference capability is improved, and the map consistency and reliability of long-term routing inspection are ensured.

The invention relates to a slam loopback detection method based on vision and laser line feature fusion. The method comprises the following steps: S1, extracting three-dimensional line features from laser radar point cloud data; s2, extracting two-dimensional line features from the visual image, and constructing an image line feature dictionary; s3, projecting the three-dimensional laser line features extracted in S1 to an image, and performing angle difference and Hausdorff distance-based matching with line segments in an image line feature dictionary to generate fusion line features; s4, in a loopback detection stage, through geometric error minimization and semantic consistency verification, determining that loopback of the current frame and the historical key frame is closed; and meanwhile, the fusion line features are added into the pose image in real time to optimize the current pose. According to the invention, the precision, robustness and long-term stability of the SLAM system in the mine environment are improved, and reliable technical support is provided for unmanned transportation of mines.

This invention belongs to the field of target detection technology, specifically relating to a multimodal feature fusion method for water target detection. The method includes the following steps in sequence: a preprocessing step, including: based on two adjacent RGB image frames as input, separating mirror perturbations to generate a mirror mask and a motion cue map; an alignment and fusion step, including: using a stable frame as the primary modality and the mirror mask and motion cue map as secondary modalities, extracting primary and secondary modal features; and inputting the fused features into a detection head to obtain initial candidate boxes and category scores; and a loop closure determination step, including: based on the conformal saliency ratio calculated for the initial candidate boxes and calibrated within the same frame, outputting the final detection result for that frame. This invention can automatically identify and eliminate unstable candidates caused by reflection. It also enhances the robustness and consistency of the model in complex scenes such as strong reflections, water wave disturbances, and changes in viewing angle.

This invention discloses a laser SLAM loop closure detection method based on gridded uniform sampling and multi-resolution registration, comprising: discretizing and constraining historical keyframes through spatial gridding organization and structural feature extraction to generate a first candidate loop closure set with structural consistency; performing initial pose correction on candidate loops in the first candidate loop closure set based on multi-source pose information, and performing stepwise optimization matching of candidate loops in combination with multi-scale point cloud representation to obtain a second candidate loop closure set; constructing a loop closure quality assessment model based on the point cloud matching results of candidate loops in the second candidate loop closure set, and filtering candidate loops through spatial distance-based matching score calculation and adaptive threshold determination mechanism, combined with geometric consistency verification, to obtain a final loop closure set.

This invention discloses a multi-sensor fusion SLAM method for dynamic scenes, belonging to the field of robot perception. The method includes the following steps: preprocessing visual, laser, and IMU data, including time and spatial synchronization; establishing sensor mathematical models and simultaneously constructing a visual-inertial subsystem and a laser-inertial subsystem, which are fused in a tightly coupled manner. The laser-inertial subsystem assists the visual-inertial subsystem in scale recovery, and the visual-inertial subsystem assists the laser-inertial subsystem in loop closure detection. Finally, a globally nonlinear optimization factor graph model is established through graph optimization, adding laser odometry factors, IMU pre-integration factors, visual factors, and loop closure detection factors to the factor graph to obtain globally optimal pose estimation and a 3D map. In dynamic environments, this invention reduces the impact of dynamic objects on localization by fusing information from multiple sensors, resulting in higher localization accuracy and robustness.

The disclosed embodiments relate to a loop detection method, an apparatus, an electronic device and a medium, where the method includes: obtaining a frame image set of a scanned object; each of frame images in the frame image set has a plurality of scanned marking points; generating an initial frame of the scanned object according to the frame image set; obtaining a candidate frame point set of each initial frame point in the initial frame based on the coordinate position information of each initial frame point in the initial frame; identifying a target frame point in the candidate frame point set of each initial frame point that meets a preset loop detection condition, establishing a link relationship between each initial frame point and a corresponding target frame point, performing a global optimization processing on the marking points of the initial frame based on the link relationship between the initial frame point and the corresponding target frame point, and obtaining a first target frame. The above technical solution can improve the accuracy and reliability of loop detection, thereby making the model generated by the final three-dimensional reconstruction more accurate and reliable.

The invention belongs to the technical field of positioning and mapping (SLAM), and discloses a visual inertial odometer method for a dark light environment. Performing real-time image enhancement on the image through an image enhancement module; a deep feature extraction module extracts feature points from the image after image enhancement, and local feature matching based on deep learning is carried out; the open vocabulary scene recognition module performs open vocabulary semantic extraction and target detection based on text prompt, and constructs a two-dimensional semantic topological graph according to scene semantic information; the mixed architecture state estimation module takes a filtered pose result as an initial value of optimization in a world coordinate system; a local mapping thread and a loopback detection thread are synchronously started, the local mapping thread is responsible for performing real-time optimization and maintenance on a map of a current local area and transmitting a key frame to the loopback detection thread, scene re-recognition is performed on the loopback detection thread based on a two-dimensional semantic topological graph constructed by an open vocabulary scene recognition module, and the scene re-recognition is performed on the basis of the two-dimensional semantic topological graph; and the closed loop detection robustness in a complex environment is improved.

The invention relates to a memory efficient visual SLAM method and system based on sparse prior embedding and map rarefaction, and belongs to the technical field of industrial robot positioning and mapping. The method comprises the following steps: extracting key frames in image frame data through a visual odometer, and constructing a key frame set; eliminating redundant key frames by adopting an information matrix retention strategy and an edge residual retention mechanism; then map point quality information is evaluated based on a feature point parallax score and a descriptor aggregation degree score index, and the contribution degree of an observed map point is comprehensively evaluated in a nonlinear function weighted fusion mode; and finally, combining grid discrete constraint and regularization in a sliding window, and realizing sparse selection and dynamic retention of map points through high-score rewards. And performing enhanced loopback detection on the basis of the constructed sparse image, and executing image feature matching and loopback confirmation by constructing a common-view region and extracting information compression features.

Techniques are provided for map-based SLAM (Simultaneous Localization and Mapping). The method includes obtaining a new externally perceived measurement from the device. New externally perceived measurements are added to the global map. The external perceived measurements produce new nodes and new edges in the global map. The method includes identifying at least one closed loop of new nodes and new edges in the global map. The method includes combining the identified at least one closed loop with closed loops previously detected in the global map, thereby producing a compressed global map. The method includes obtaining an edge weight of an edge of at least one closed loop in the compressed global map. The method includes updating the global map through new inner points and new outer points identified using edge weights. The edges classified as inner points are added to a global map. Edges classified as outer points are removed from the global map.

The invention discloses a robust real-time SLAM (simultaneous localization and mapping) method and system based on multi-modal deep learning and optimization fusion, and relates to the field of computer vision and the technical field of simultaneous localization and mapping. Aiming at the problems that an existing visual inertia SLAM system is poor in robustness, ineffective in feature matching and low in closed-loop detection efficiency in a high-speed motion, low-illumination and high-dynamic range scene, the invention provides a multi-sensor fusion framework, and the performance is improved through deep learning and a dynamic adaptive strategy by combining an event camera, a monocular camera and an IMU (Inertial Measurement Unit). The method comprises the following steps: acquiring related data on an SLAM system; synthesizing a virtual event frame image; constructing IMU pre-integration, and aligning IMU data with the fixed frame rate image; synthesizing compensation image frames; judging whether the compensation image frame synthesized at the current moment is a key frame or not, and if yes, storing the compensation image frame into a sliding window; selecting a reference frame, and performing joint initialization; and performing hierarchical optimization on the initialized SLAM system based on a sliding window and a feature extraction tool.

This invention relates to the field of computer software and game technology, specifically disclosing a method for differentiated game version updates. The method includes: collecting player operation sequences and state feedback sequences; extracting reflexive loop topology; dividing frequency bands according to operation intervals to extract group delay feature spectrum and dispersion feature spectrum; calculating loop inertia weights; and constructing a personalized reflexive tolerance baseline. The version content is decoupled into atomic difference units; group delay offset and dispersion increment are evaluated; compatible units are selected and assembled with reusable residual files to form group delay compatible slices; and verification anchor points are established. Slices are progressively injected and group delay drift is monitored. In the event of multi-player conflicts, shared reflexive arbitration is triggered, loop closure compensation is performed, and residual indexes are archived. This invention solves the technical problem of existing updates neglecting the consistency differences in player operation frequency response and the consistency of multi-player scene states, achieving synergistic optimization of player-level fine-grained differentiated updates and consistency assurance in multi-player shared scenes.

Embodiments of the present application provide a closed loop detection method and system, a medium, equipment and a program product. The method comprises: in the case that a spatial relationship between a current key frame and at least one historical key frame satisfies a preset spatial constraint condition, and a time relationship between the current key frame and an execution time of a last closed loop detection process satisfies a preset time constraint condition, performing a closed loop detection process on the current key frame and the historical key frame to generate a closed loop detection result; the closed loop detection result is used to represent that the current key frame and the at least one historical key frame constitute a closed loop relationship or do not constitute a closed loop relationship. Since the closed loop detection process is performed on the candidate historical key frame only when the two preconditions of the preset spatial constraint condition and the time constraint condition are satisfied, the candidate frame is effectively screened in time and space before entering the subsequent related process. Therefore, the method can reduce the occurrence rate of closed loop false detection in a complex scene.