33results about How to "Positioning results are stable" patented technology

The invention relates to an unmanned vehicle navigation and positioning precision correction method based on GNSS and visual odometer fusion. The unmanned vehicle navigation and positioning precision correction method comprises the steps of: (1) detecting lane lines based on monocular vision; (2) optimizing positioning precision of a monocular visual odometer; (3) and correcting positioning precision of a GNSS/VO integrated navigation system. The unmanned vehicle navigation and positioning precision correction method utilizes the lane lines for assisting the visual odometer, then fusing the visual odometer with GNSS positioning to improve the reliability of a vehicle positioning system; the GNSS/VO has a high complementary feature, the GNSS can obtain a long-term stable positioning result, and VO can obtain short-term high-precision positioning data, two sensors are used for filtering a difference value of position measured values, and estimated value of VO system errors is used for correcting errors of the VO system, thereby realizing the purpose of restricting long-term drift of the VO data by utilizing the GNSS data.

The invention provides a lane line positioning method based on multi-feature fusion and a polymorphism mean value. According to the method, images of road conditions in front of vehicles are obtained through a camera; the gray level distribution information of lane lines and the color distribution information in an HSI space are fused according to weight to obtain a lane line suspected area according to the pavement gray scale normal distribution features and the HIS color space information; then, according to priori knowledge, the searching is carried out from the middle of a road to the two sides for excluding non-lane regions, and sub images which avoid the non-road feature interference and contain the lane lines are obtained; then, the sub images are subjected to margin determination; Hough transformation is adopted for detecting straight lines for determining the positions of the lane lines in the current frame; and finally, through fusing the multi-frame lane line states and utilizing the actual condition that the lane line positions cannot suddenly change, the mean value of detection results of continuous frames is worked out, and stable lane line positioning results are obtained.

The invention discloses a supercompact combination satellite navigation receiver which comprises a radio frequency front end, a signal preprocessor, a base band prefilter, a navigation filter, a micro inertial navigation module, a pseudo-range and pseudo-range rate calculating module and an ephemeris resolving and start selection calculating module. A self-adaptive filter is arranged between the base band prefilter and the navigation filter. An output of the base band prefilter is used as an input to be switched in the self-adaptive filter and then an output of the self-adaptive filter is used as an observed quantity to be input into the navigation filter to carry out data fusion, and thus, the stability and the tracking accuracy of a tracking loop can be obviously improved, so that the adaption capacity of GNSS (Global Navigation Satellite System) positioning on the environment and the motion state is improved. Furthermore, the vector tracking loop is formed by the signal proprocessor, the base band prefilter, the self-adaptive filter, the navigation filter and the pseudo-range and pseudo-range rate calculating module; and the navigation filter adopts a federated filter, so that the stability and the tracking accuracy of the tracking loop can be improved, and thus, the performances of reliability, positioning accuracy and the like of the GNSS positioning are improved.

The invention discloses a multi-direction collection combined clustering WiFi fingerprint indoor positioning method. The method comprises the steps that (1), a signal collector selects east, west, south and north directions for signal strength value collection; (2), signal strength values in the (1) are denoised; (3), fingerprint data is clustered, indexes are established, and after the final element is screened completely, signal strength in each location fingerprint is sorted according to a descending order; and (4), classes which have H maximum values the same as those of to-be-measured point signals are found according to a set of each cluster in the (3), and coordinate locations of the to-be-measured points are computed through adoption of K-nearest neighbor. According to the method,a multi-direction collection method and a denoising clustering method are employed for fingerprint library production. Compared with a traditional fingerprint positioning method, the method has the advantages that the positioning precision is improved by 15%, the positioning time is reduced by 20%, and the defect that in a traditional WiFi fingerprint positioning process, the data processing is not reasonable enough, the fingerprint library reliability is low and the positioning timeliness is low is effectively solved.

The invention discloses an indoor positioning method based on WiFi, Bluetooth and pedestrian dead reckoning (PDR) fusion, and the method is characterized in that the method comprises the following steps: 1) dividing a positioning region, and building a WiFi offline fingerprint data 2) clustering the training samples to obtain a centroid-like vector; 3) adopting a hierarchical clustering algorithm to obtain positioning coordinates; (4) achieving Bluetooth positioning by adopting an improved weighted centroid algorithm, (5) obtaining a WiFi and Bluetooth fusion positioning coordinate by fusing WiFi and Bluetooth positioning, and (6) obtaining a final positioning coordinate in combination with PDR. The method has the characteristics of high positioning precision and low software calculation amount, and can meet the real-time positioning precision requirement.

The invention discloses a signal source positioning method for uniform velocity fields. The method comprises the following steps: firstly constructing hyperboloid equations of a signal source position by using a residual equation of the theoretical arrival time and observation arrival time of two sensor pairs; constructing a virtual field in a local coordinate system according to each hyperboloid equation; superposing and averaging all the hyperboloid equations in a space by utilizing a coordinate conversion technology so as to obtain an integral virtual field; and searching the position of the maximum value of the integral virtual field in the space by utilizing a common unconstrained optimization algorithm and taking the position as the signal source positioning result. The objective function of the integral virtual field has continuity and differentiability in the space, so that the positioning method has good resistance to the severe picking errors, and has the characteristics of being high in positioning speed and good in stability.

The present invention discloses an indoor and outdoor integrated positioning and navigation method and system, and relates to the field of positioning fusion. The method comprises: step 1: obtaining GPS, Bluetooth and inertial positioning data; step 2: after performing adaptive weighting, Kalman filtering and distance-based correction on the positioning data to complete positioning solving to obtain a final positioning point; step 3: performing point set division on a navigation line to obtain a reference point for projection calculation, and using a sliding window and curve geometric projection to project the final positioning point onto the reference point of the navigation line to obtain a projection point; and step 4: using the projection point for navigation and map display. Accordingto the technical scheme of the present invention, an adaptive weighting and Kalman filtering model is used to obtain a smooth positioning result, and distance-based correction is performed on the result to obtain stable and accurate positioning, so that the difficulties of the calculation amount, the real-time performance and stability caused by the fusion of inertial positioning, Bluetooth positioning and GPS positioning are overcome, and smooth indoor and outdoor integrated positioning and navigation can be achieved.

The invention discloses a WLAN indoor positioning method based on sample capacity optimization and variance analysis. The method comprises the steps of firstly during an offline stage, based on analysis on RSS sample characteristics, using an OC function to optimize a fingerprint database to construct an allowable minimum RSS sample; then, selecting an optimal AP (access point) for constructing the fingerprint database via an AP selection method based on information gain; and at last, roughly positioning a target terminal via a statistical hypothesis variance analysis method by using the affect of the RSS sample mean and sample capacity on similarity between samples during on online stage, and thus providing a KNN algorithm based on variance analysis to finely position the target terminal.According to the method provided by the invention, the problem of influence of the blindness and unreliability of fingerprint data collection and the fluctuation of the RSS sample on the positioningperformance is solved. Experimental results show that by using the method, the relatively stable positioning result is acquired with the limited sample capacity, and the labor and time overheads are significantly reduced.

The invention discloses a positioning method for a Sagnac distributed optical fiber sensing system based on convolutional neural network ensemble learning. The method comprises the steps of taking a fixed interval point on a sensing optical fiber as a disturbance point, enabling the sensing system to obtain an interference signal generated by the simulation disturbance of each point, carrying outthe preprocessing of the interference signal, and enabling the preprocessed interference signal to serve as a training set and a verification set, training two convolutional neural network models fordifferent loss functions to enable the two CNN models to accurately position near-end and far-end disturbances respectively, combining training results of the two models through an ensemble learning method to obtain a disturbance position prediction model based on CNN ensemble learning, and optimizing parameters of each model through the verification set. A to-be-positioned interference signal ispre-processed as a test sample, and the to-be-positioned interference signal is tested by using the trained prediction model to obtain a disturbance position of the to-be-positioned interference signal. The method does not need signal demodulation, is low in system complexity, is not sensitive to noise, is simple in data processing method, is stable and accurate in positioning result, and can be used for the disturbance positioning of an annular or linear Sagnac distribution optical fiber sensing system.

The invention provides an autonomous parking navigation method based on parking lot map data guidance, and belongs to the field of autonomous parking, and the autonomous parking navigation method comprises the steps: building a parking lot high-precision map containing parking lot environment information, obtaining the position of a vehicle through the comparison and matching of the obtained vehicle surrounding perception information and the parking lot environment information, selecting a preset parking space from the parking lot high-precision map, so that the system plans the shortest driving path between the vehicle and the selected parking space according to the position of the selected parking space and the position of the vehicle, and the vehicle is enabled to rapidly drive to the selected parking space; meanwhile, in the manual driving state, the vehicle surrounding information is extracted and compared with the parking lot environment information, when a unique matching result does not exist, the vehicle surrounding information is collected, map information is generated and fused with a parking lot high-precision map to generate a new parking lot high-precision map, and therefore secondary parking navigation of the vehicle and parking navigation of other vehicles are facilitated.

The invention discloses a device for indoor three-dimensional space positioning based on rotary light patterns. The device comprises a periodical flashing light source and a uniform rotation special-shaped fan blade, wherein the plane of the fan blade is parallel with ground; a connecting line between the center of the light source and the center of the fan blade is vertical to the ground; a plurality of devices are arranged on the plane which is parallel with the ground, and the fan blade rotary cycles and light source flashing cycles of the devices are different. The invention also provides a method for the indoor three-dimensional space positioning based on the device. The method comprises the following steps of detecting the change rule of different light patterns by a photosensitive sensor, so as to judge an included angle between a to-be-positioned target and the rotation axis vertical line of the fan blade, and a direction included angle between the to-be-positioned target and the centerline of each fan blade; then, determining the position of the to-be-positioned target by the position of the light source. The device has the advantages that the positioning accuracy is high, the positioning result is stable, the layout cost is low, the hardware requirement of target positioning is low, the number of simultaneously positioned targets is more, the expanding of positioning range is easy, and the like.

The invention discloses a robot positioning control method and system based on point cloud normal vector direction clustering servo, and belongs to the technical field of visual servo, and the robot has six-degree-of-freedom motion and is fixedly connected with an RGBD depth camera; the method comprises the following steps: acquiring an ideal point cloud image of a to-be-positioned target; calculating and clustering normal vectors of all points in the ideal point cloud image, and obtaining a clustering centroid set to obtain an actual point cloud image of the to-be-positioned target; calculating and clustering normal vectors of all points in the actual point cloud image, and obtaining a clustering centroid set MC; calculating a rotation matrix and a translation transformation vector of theRGBD depth camera at the current position relative to the ideal position according to the clustering centroid set and the MC, and calculating the speed VC of the RGBD depth camera based on the calculation result; and if the VC is not converged to 0, enabling the robot to move according to the VC for a preset time, recalculating the VC and judging until the VC is converged to 0. The invention hasa stable and high-precision positioning result for a weak texture target, and has better adaptability.