442 results about "Tracking model" patented technology

The invention discloses an intelligent question bank retrieval and recommendation system based on an artificial intelligence knowledge graph, and relates to the technical field of artificial intelligence education. Comprising a knowledge graph construction module which is used for processing original education data and constructing a neighborhood knowledge graph comprising a hard preposition relation and a soft incidence relation; the user knowledge state graph construction module is used for constructing a user personal knowledge state graph isomorphic to the domain knowledge graph, and dynamically calculating a mastery index of each knowledge node through a deep knowledge tracking model based on user historical answer data; according to the method, by constructing the domain knowledge graph containing the hard preposition relation and the soft incidence relation, discrete knowledge points are organized into the structured network conforming to the cognitive law, so that the system can understand and follow the internal logic between knowledge, and a learning path which is clear in organization, efficient and coherent is generated.

The invention provides a multi-mode remote sensing target tracking and positioning and intention discrimination method and device. The method comprises the following steps: acquiring a plurality of visible light image frames and a plurality of infrared light image frames, and carrying out frame alignment operation on each visible light image frame and each infrared light image frame to obtain a plurality of groups of effective image frame pairs; for each group of effective image frame pairs, determining tracking identification information of each detection object in the effective image frame pairs based on the effective image frame pairs and a pre-trained multi-modal detection tracking model; for each detection object, determining longitude and latitude tracks of the detection object based on the tracking identification information and a back projection mapping function; and determining the behavior intention of each detection object based on a behavior recognition model and the longitude and latitude tracks of each detection object. The accuracy of target tracking and behavior intention recognition in the remote sensing video can be improved.

The invention relates to a big language model enhanced knowledge adaptive learning planning system, and belongs to the field of intelligent education. The system comprises a knowledge center module, a learner portrait module, a path planning module and an intelligent learning guiding module, and the knowledge center module extracts entities and relationships from a multi-modal data source by using a large language model to construct a knowledge graph; the learner portrait module collects multi-dimensional learning data of the user and maps the multi-dimensional learning data to corresponding nodes of a knowledge graph, and dynamically deduces a learner portrait through a Bayesian knowledge tracking model; the path planning module generates an initial learning path based on the knowledge graph and the learner portrait, establishes a collaborative filtering analysis model, predicts and optimizes the expected effect of the current learner following the initial learning path in combination with a Bayesian knowledge tracking model, and finally generates a target learning path; and the intelligent learning guiding module generates a standardized knowledge card for each knowledge node on the target learning path through a security retrieval enhancement generation technology.

The invention aims to provide a SiamFC-pytorch tracking model pre-processing and post-processing acceleration method and a SiamFC-pytorch tracking model pre-processing and post-processing acceleration system based on an NPU (Network Processing Unit). The system is realized based on a Yulong810SOC, and comprises a video input module (1), a hardware decoding module (2), a detection module (3), a tracking module (4) and a turntable control module (5), the method comprises the following steps: a, a video input module receives a video stream, and the video stream is decoded into an original image in a YUV format through a hardware decoding module; b, the detection module runs a YOLOv5 model through a neural network engine of the NPU; c, an NPU pre-processing and post-processing unit of the tracking module executes template branch pre-processing, search branch pre-processing and search branch post-processing on the YUV image and the target initial bounding box; and d, the turntable control module adjusts the posture of the camera turntable according to the real-time coordinates of the target. The method is applied to the technical field of computer vision and artificial intelligence hardware acceleration.

The invention belongs to the field of explosion load and response prediction, and particularly relates to an impact load and response intelligent prediction system and method based on machine vision. An impact load and response intelligent prediction system based on machine vision comprises an image data preparation module, an image recognition and segmentation module, a numerical calculation module and a load and response prediction module, and the image data preparation module obtains, marks and enhances an image data set required by a target detection and tracking model; the image recognition and segmentation module recognizes and segments splashing fragments and dynamically tracks the splashing fragments; the numerical calculation module generates a sample database required for training the machine learning model; and the load and response prediction module outputs a forward prediction model, a reverse prediction model and a prediction result. According to the prediction system and method, computer vision, a numerical calculation method and machine learning are fused together, and the problems that a traditional method is low in efficiency and cannot conduct reverse prediction according to fragment distribution after impact explosion are solved.

The invention discloses a spraying control method of a waveform guardrail spraying manipulator, which relates to the technical field of industrial automation and robot control, and comprises the following steps: acquiring relative position change data of the manipulator and the surface of a waveform guardrail in a rotation process through a pre-established manipulator rotation track model; and for each time node, recording the spraying angle change and distance dynamic adjustment information, establishing a dynamic space mapping relationship under a manipulator rotation state for spraying control, and triggering a secondary spraying supplementing control instruction according to the coating quality consistency feedback, so that the reliability and stability of the spraying supplementing effect can be further ensured, and the spraying efficiency is improved. Therefore, the equipment utilization rate of the manipulator and the automation and intelligence level of spraying operation are remarkably improved while the uniformity of the coating and the consistency of spraying quality are improved.

The invention relates to the technical field of artificial intelligence, and discloses a Transformer training method for paint surface AI defect identification, which comprises the following steps: synchronously acquiring RGB images and three-dimensional point cloud data through a six-axis mechanical arm to form a multi-modal data set; generating synthetic defect data by using fluid dynamics and a ray tracing model to expand a sample set; realizing feature alignment by adopting cross-modal contrast learning; aggregating defect structure capsules through a dynamic routing attention network; performing knowledge retrieval and fusion in combination with the prototype memory matrix; and finally, the sorting mechanical arm is driven to execute sorting or rechecking operation based on the multi-task output and the uncertainty score. According to the method, the problems of insufficient structural representation of complex defects, weak generalization ability of rare defects and low reliability of sorting decision are solved, and high-precision and high-reliability automatic defect detection and sorting are realized.

The invention discloses a robot dog target following method based on computer vision. The robot dog target following method comprises the following steps: S10, preparing human body data, human body weight identification data and a single target tracking data set; s20, designing a lightweight human body detection network architecture and a detection human body frame, and performing model training based on the human body data; s30, constructing a lightweight single-target tracking model, selecting a detection frame of a to-be-tracked target as initialization input, receiving an initial frame transmitted by a robot dog camera as a template image, and performing specified template tracking and following; s40, constructing a lightweight full-scale model, starting a relocation mechanism after a lost target is tracked by a single target, and controlling the robot dog to turn around in situ, globally search a specified target and relocate a disappearing target; and S50, based on the tracking result of the single target, converting the target position on the image to an actual three-dimensional coordinate, and controlling the motion displacement of the robot dog to complete the following operation.

The invention relates to a knowledge tracking model research method integrating difficulty perception and memory enhancement. According to the method, a graph attention network coding exercise-knowledge point topological relation is constructed, and a difficulty embedding layer and a gating fusion mechanism are combined to realize joint characterization of difficulty features and a topological structure. The difficulty coefficient is innovatively introduced as a dynamic regulation factor of attention weight in memory network updating, the memory intensity of high-difficulty exercises is enhanced, and the discrimination and topological consistency of knowledge representation are improved through a multi-task optimization framework integrating graph structure loss and contrast loss. According to the method, challenges of exercise difficulty perception and knowledge point topological relation modeling are effectively solved, experimental results show that the AUC of the model reaches 0.93 (improved by 7.8% compared with traditional BKT) on an ASSIST2009 data set, the AUC of the model reaches 0.90 (improved by 0.21 compared with KSGAN) on an EdNet data set, objective indexes and teaching scene verification show that the method has remarkable advantages in the aspects of knowledge point correlation modeling and knowledge tracking, and the method is suitable for popularization and application. The method provides an innovative solution for personalized education, and has technical breakthrough and industrial application values.

The embodiment of the invention discloses an image target labeling method and device, electronic equipment and a storage medium. According to the embodiment of the invention, a to-be-labeled video frame sequence of a road camera can be acquired; detecting any current frame in the video frame sequence based on a preset target detection model, and generating a detection frame for the target detection object; obtaining a prediction frame in the current frame, wherein the prediction frame is generated by a preset target tracking model according to the motion state of the target tracking object in the previous frame or multiple frames and the position of the detection frame; and matching the detection frame in the current frame with the prediction frame, if matching succeeds, allocating the historical identity identifier of the target tracking object corresponding to the prediction frame to the target detection object corresponding to the detection frame, and if matching fails, allocating a new identity identifier to the target detection object corresponding to the detection frame. Therefore, based on the time-space coherence of the video, the target is tracked and labeled efficiently and accurately.

The invention discloses a target tracking model based on an enhanced target detection algorithm, and relates to the technical field of real-time target tracking. A CD-YOLO detector is constructed; a robust multi-target tracker is constructed; constructing a collaborative detection-tracking optimization mechanism; evaluating and optimizing the performance; according to the method, a deformable convolution and coordinate attention mechanism is introduced into YOLOv10, so that the perceptual ability of the model to geometric deformation and spatial information is enhanced; the CD-YOLO and the enhanced StrongSORT tracker are combined, so that the detection precision and the tracking stability of the system in complex scenes such as dense crowds, serious shielding and variable visual angles are remarkably improved; a collaborative optimization mechanism between detection and tracking is established, so that a detection result and tracking feedback can be mutually enhanced, a closed-loop robust tracking process is formed, and the overall precision and stability are remarkably improved on the premise that the real-time performance of the system is not affected.

The invention relates to the technical field of vehicle intelligent driving and active safety control, and discloses an active obstacle avoidance method and system considering multi-source uncertainty and fault positioning, and the method comprises the following steps: S1, predicting the tracks of a normal vehicle and a sideslip vehicle, and obtaining the predicted position and covariance of the normal vehicle and the sideslip vehicle; s2, constructing an extended state equation containing a positioning fault, and estimating and compensating a positioning deviation; s3, establishing a linear error tracking model, unifying various uncertainties into equivalent disturbance, and solving a disturbance invariant set; s4, generating a candidate obstacle avoidance track based on the compensated state; s5, track comfort is evaluated, hard constraints are checked, and infeasible tracks are removed; s6, transversely expanding the contour of the vehicle by using a disturbance invariant set, and carrying out longitudinal and transverse separation expansion on the contour of the target vehicle based on the predicted covariance; s7, performing robust collision detection by using the disconnect axis theorem; and S8, carrying out robust tracking control by adopting Tube-MPC. According to the invention, safe and stable obstacle avoidance control can be realized under complex disturbance and fault positioning conditions.

The invention discloses an intelligent obstacle recognition method for an unmanned loader, and aims to solve the problems of low obstacle recognition precision, poor real-time performance and insufficient coordination of obstacle avoidance and operation in a complex operation scene. The method comprises the following steps: establishing a multi-source sensing system such as a laser radar and a camera, and realizing data time sequence synchronization, space unification and noise optimization through FPGA hardware; obstacle space and category features are extracted through double-branch parallel detection, and a modal weight fusion result is dynamically distributed in combination with a scene; an improved Hungary algorithm and a classification tracking model are adopted to realize obstacle ID stable association and trajectory tracking, and trajectory prediction is completed based on LSTM and Kalman filtering; and a dynamic avoidance area is constructed, and an obstacle avoidance path is generated through global-local bilevel programming and a safety potential field algorithm and is corrected in real time. The obstacle recognition robustness and real-time performance in a complex environment are improved, and safe and efficient operation of the unmanned loader is guaranteed.

The invention discloses an ISAC angular domain energy imaging method and an ISAC angular domain energy imaging system based on actual measurement channel response calibration ray tracing, and mainly solves the problem that an imaging model faces serious channel mismatch and cross-scene generalization limitation due to the fact that simulation data lacks physical consistency constraint in the prior art. According to the scheme, the method comprises the steps that geometric features are extracted from public map data or laser point cloud data, and an ISAC scene environment geometric model is constructed; acquiring an actual measurement channel impulse response containing multipath information by using the communication and sensing integrated equipment; performing path matching on the actual measurement channel impulse response and the ray tracing simulation channel response, inverting environment propagation parameters and correcting scattering parameters; and inputting the calibrated parameters into a ray tracking model, counting energy contribution in each space direction under the condition of a single antenna, obtaining angular domain energy distribution and confidence coefficient, and mapping the angular domain energy distribution and confidence coefficient to a spherical grid to form an imaging result. According to the method, physical mismatch between simulation and real environments is effectively eliminated, low-cost and high-precision angular domain energy imaging and confidence evaluation are realized under the condition of a single antenna, and the method can be used for imaging reconstruction of an ISAC system, environmental perception and generation of a neural network training data set.

A high-performance loosely coupled multi-modal data fusion system for an intelligent driving environment perception system, and on-board equipment are disclosed in the present disclosure. The data fusion system includes a fusion detection model based on a modal-specific feature interaction strategy, configured to convert LiDAR point clouds, camera images, and millimeter-wave radar point clouds into unified bird's-eye view (BEV) features and perform multi-modal fusion; and a fusion tracking model based on a cascade coupling data association strategy of motion-appearance features, configured to perform subsequent trajectory tracking and matching based on feature information of the multi-modal fusion. A VoD dataset and a K-Radar dataset are selected for training, validating, and testing comprehensive performance of the models. An inference model is accelerated by applying TensorRT, to be quantified and deployed on an on-board computing test platform.

The invention belongs to the technical field of mine geological exploration, and discloses a strip mine geological environment simulation early warning method and system based on a dynamic evolution model. The method comprises the following steps: constructing a dynamic simulation evolution model based on multi-source data information of an exposed mine in a preset range; based on the dynamic simulation evolution model, geological environment evolution simulation and estimation are carried out on an area in a preset range, an area where an abnormal geological environment possibly exists is identified, and monitoring equipment is arranged in the area where the abnormal geological environment possibly exists; the method comprises the following steps: collecting monitoring data in a preset range through monitoring equipment, and processing and analyzing the monitoring data to obtain a data analysis result; and constructing a geological environment data tracking model based on the data analysis result, carrying out dynamic tracking on the geological environment by using the geological environment data tracking model, and generating early warning information according to the tracking result. According to the scheme, the defects that a traditional exploration method is single in data and lagged in judgment can be effectively overcome, and the mining safety risk and monitoring cost of the strip mine are greatly reduced.

The invention discloses an unmanned aerial vehicle relay tracking system and method for cloud side-end cooperation in an urban scene, and belongs to the technical field of intelligent traffic and unmanned aerial vehicle cooperative control. In order to solve the problems of insufficient endurance and limited sight distance of existing single-machine tracking in an urban large-range scene, a physically distributed and logically unified cloud-side-end three-level architecture is constructed, wherein a cloud end is responsible for natural language instruction analysis and global road network monitoring; the edge end serves as a regional scheduling hub and is responsible for operating a high-precision detection model, extracting a target Re-ID feature fingerprint and calculating a cross-machine relay time window; and the terminal unmanned aerial vehicle only operates a lightweight tracking model and a visual servo control law. According to the method, the urban area is divided into a plurality of honeycomb grids, seamless locking and relay of a target among unmanned aerial vehicles in different grid responsibility areas are realized by using a cross-view feature relay protocol and a collaborative finite-state machine, and the engineering problem of long-time-sequence and large-range vehicle tracking is effectively solved.

The invention provides a tracking method of dialogue-related knowledge points and electronic equipment. According to the embodiment, the knowledge points related to the dialogue are predicted through multiple methods such as the dialogue knowledge point prediction model, the retrieval enhancement generation method and the large language model LLM hierarchical prompt method, then the knowledge points predicted through the methods are fused to track the knowledge points related to the dialogue, the deviation of a single method is reduced, and the confidence coefficient of dialogue tracking is also improved; according to the embodiment of the invention, an autoregressive target knowledge tracking model is trained by means of the fusion result of the knowledge point set of each dialogue, the historical dialogues before each dialogue and the problems in each dialogue, so that the change of knowledge points in multiple rounds of dialogues can be timely captured when the target knowledge tracking model is subsequently utilized to perform knowledge tracking on the dialogues; and the knowledge tracking accuracy is improved.

The invention relates to the technical field of knowledge tracking, and discloses a graph enhanced double-memory collaborative knowledge tracking model based on an ACT-R cognitive architecture. Comprising a static knowledge structure coding module based on hypergraph projection, a batch-level dynamic learning track construction and coding module, a cross-graph gating fusion mechanism, a sequence modeling module and an expert hybrid prediction module. According to the method, long-term stable structured semantic association between concepts in declarative memory is modeled through a static knowledge structure diagram, a dynamic learning trajectory diagram based on batch reconstruction is designed to accurately capture an evolution rule of a behavior sequence in programmed memory, and on the basis, a cross-diagram gating fusion mechanism and a hybrid expert mechanism are introduced, so that the evolution rule of the behavior sequence in the programmed memory is accurately captured. And self-adaptive fusion and multi-path decision of double-graph features are realized.

The invention discloses an intelligent teaching assisting method and system integrated with the whole process and total elements of education and teaching, and relates to the field of education and teaching. According to the method, misunderstanding concept classification models are integrated, and a dynamic knowledge graph containing target subject knowledge is established; clustering knowledge concepts in the dynamic knowledge graph by adopting a Mapper algorithm of topological data analysis to obtain a course map; when the user completes interaction of the selected theme cluster, a cognitive state matrix is obtained by adopting a graph knowledge tracking model, an emotion category probability distribution vector is obtained by adopting an emotion classification model, and a learner state vector is obtained; a large language model optimized through process supervision and reinforcement learning is adopted as an inference engine; and outputting a targeted teaching strategy and teaching content by using an inference engine according to the learner state vector. According to the method, a personalized teaching environment which can perform smooth and dynamic natural language interaction and can accurately diagnose and effectively correct the specific deep-level cognition mistake of students in a target subject can be created.

The invention provides a staged forgetting rate-based student knowledge mastering prediction method and related device, and the method comprises the steps: firstly obtaining education platform log answer data, carrying out the standardization processing of the data, obtaining a standardized answer data set, and determining the stage affiliation result of each student based on the data set through employing a Bayesian probability inference method, meanwhile, a student answering time sequence is constructed, context distance data between questions in the sequence is calculated, then a staged forgetting rate knowledge tracking model is constructed in combination with a student stage attribution result, the answering time sequence and the context distance data, training is completed, and finally the trained model is adopted to predict the knowledge mastering condition of the student. By matching the corresponding forgetting rate parameters for the students in different learning stages, the knowledge forgetting characteristics of the students in different stages can be accurately captured, the prediction precision of the knowledge mastering state of the students is effectively improved, and the prediction result is more fit for the actual learning level of the students.

The invention relates to the technical field of identity recognition, in particular to an identity recognition method and system based on multi-modal data. The invention provides an identity recognition method based on multi-modal data, which adopts a user use dimension, a user interaction dimension and a user voiceprint dimension, thoroughly avoids legal compliance risks caused by a face recognition technology, constructs a non-biological recognition system based on sensor motion trail capture, touch screen interaction behavior modeling and voiceprint feature extraction, and improves the recognition efficiency. Therefore, all the feature data cannot restore the original biological attributes, and the hidden danger of privacy litigation is eliminated from the source. Meanwhile, depending on a three-dimensional dynamic fusion mechanism of the sensor, the touch screen and the voiceprint, full-scene precision jump, compensation of the dark light defect by the motion inertia of the sensor capture equipment, noise interference resistance by a touch screen track modeling fine operation habit and physiological uniqueness abstract representation of the voiceprint spectrum are realized, and the three forms a complementary enhancement effect through dynamic weights.

The application belongs to the technical field of personalized learning, and discloses a multi-level intelligent cognitive tracking method, system, storable medium and terminal, the method comprising: introducing Bloom cognitive domain education target classification, constructing test question knowledge cognitive tensor TKC, collecting learning resources and answer data of learners, and generating a sequence of learner time sequence answer pairs; introducing a multi-attribute cognitive diagnosis method, combining a deep neural network, and constructing a cognitive level mining model; sorting and encoding the cognitive level mining results of the learners to obtain deep representation features, combining a self-attention mechanism, constructing a multi-level intelligent cognitive tracking model, and further predicting the answer performance of the learners on the test questions. The application is beneficial to accurately and finely modeling the overall knowledge structure and specific level of the learners, thereby promoting personalized learning of the learners and providing a new idea for mining and tracking the cognitive state and level of the learners in an online learning platform.

The invention provides a photovoltaic power station generating capacity prediction method, and belongs to the technical field of photovoltaic power stations, and the method comprises the steps: carrying out the feature extraction of a photovoltaic module image, building a health state feature vector, inputting a multi-scale meteorological feature into a multi-scale meteorological fusion prediction model, and outputting a solar irradiance prediction sequence; and inputting the component health state feature vector into a component attenuation dynamic tracking model to output a power attenuation coefficient, calculating an initial power generation prediction value according to irradiance prediction and the attenuation coefficient, carrying out weighted fusion, and carrying out Kalman filtering smoothing processing and Box-Cox conversion equalization processing on a prediction deviation sequence to obtain a final prediction result. The technical problem that the prediction precision is insufficient due to the fact that photovoltaic power generation capacity prediction cannot give consideration to multi-time-scale meteorological changes and long-term attenuation characteristics of assemblies at the same time is solved.

The invention belongs to the technical field of computer vision and image processing, and discloses a cross-modal target tracking method based on a collaborative strategy and a related device. The cross-modal target tracking method comprises the following steps: acquiring a template image and a current frame search image; based on the obtained template image and the current frame search image, performing cross-modal target tracking by using the trained cross-modal target tracking model to obtain a cross-modal target tracking result; the cross-modal target tracking model comprises four branch networks and a multi-modal region suggestion network; each of the four branch networks comprises a feature extraction module, a multi-scale channel adaptive enhancement module and a hierarchical progressive attention fusion module. According to the technical scheme, challenges such as scale change, deformation and texture blurring can be effectively handled, and the tracking stability and robustness under the conditions of shielding, modal interference and low illumination are improved.

The invention discloses a parking state intelligent sensing method and device based on target detection tracking, and relates to the technical field of image recognition. The method comprises the following steps: deploying a vehicle target detection model and a vehicle target tracking model in a cloud server to all edge computing gateways of a parking lot monitoring system; inputting a video stream collected by the camera into a vehicle target detection model of an edge computing gateway, and carrying out vehicle target detection to obtain a plurality of vehicle targets; inputting the plurality of vehicle targets into a vehicle target tracking model of the edge computing gateway, and performing vehicle target tracking to obtain active tracks of the plurality of vehicle targets; and performing dynamic optimization by using an improved optimization algorithm according to the active tracks of the plurality of vehicle targets to obtain a parking state list of the parking lot. The problems that in the prior art, cloud processing is relied on, detection and tracking are disjointed, state perception is simple, and the model generalization ability is weak are solved.

The invention discloses a dynamic image motion correction method based on a graph neural network, and the method comprises the following steps: a), constructing a target region trajectory tracking model based on the graph neural network, and achieving the high-precision motion trajectory modeling; b) analyzing an image displacement mode in real time through a dynamic discrimination algorithm; and c) realizing adaptive image motion correction based on the target area track features. According to the method, the problem of complex motion trail modeling limitation caused by dependence on fixed template matching in a traditional method is innovatively solved, and the problem of spatial-temporal characteristic aliasing caused by unsteady state deformation of nervous tissues is effectively solved. According to the technical scheme, the dependence on a hardware synchronization signal acquisition module is eliminated, the resource configuration requirement of the edge computing equipment is remarkably reduced, and meanwhile, the multi-scale time sequence integration efficiency is improved. According to the method, dynamic imaging reconstruction of subcellular neural activities can be realized, and dynamic change details in a target neuron issuing process can be accurately restored.

The invention relates to the field of knowledge tracking, and discloses a time-guided attention and mixed expert collaborative knowledge tracking method and system, and the method comprises the steps: obtaining a historical answer interaction sequence of a student; constructing and training a knowledge tracking model; and inputting the historical answer interaction sequence into the knowledge tracking model to complete prediction of the answer condition of the student. According to the method, the individuation of knowledge tracking and the time sequence modeling capability are effectively improved through a time-guided attention and mixed expert cooperation mechanism. Wherein the time-guided attention module is combined with multi-scale forgetting bias, so that stable modeling for long-term dependence is enhanced; and the hybrid expert module dynamically activates an adaptive expert according to the learning rhythm of the student, thereby realizing adaptive capture of heterogeneous learning behaviors. The method does not need to depend on complex external semantic information, can achieve high-quality prediction only based on basic interaction data, and remarkably improves the robustness and prediction accuracy of irregular time intervals.

The invention belongs to the field of vehicle trajectory tracking control, and discloses an intelligent vehicle trajectory tracking control method and system based on differential programming, and the method comprises the steps: constructing a vehicle dynamics model; a reference trajectory time-varying curvature is introduced, delay of a steering execution mechanism is expressed as pure delay plus a first-order inertia link, and construction of a road tracking model is completed; assuming that the system is in a steady state, calculating reference values of a state quantity and a control quantity; constructing an MPC optimization problem, converting the MPC optimization problem into a QP problem, and calculating to obtain an MPC loss value; constructing a differential programming training calculation network diagram; a hypercube is used for sampling the state quantity of a vehicle, the driving speed and the time-varying reference items of a road, and construction of a network training initial state data set is completed; carrying out MPC trajectory tracking task integration, and constructing a physical model in network training; using a full-link layer network, adding tire slip angle soft constraints, completing the construction of a strategy network, carrying out end-to-end network training, and completing the control of a turning angle instruction. According to the method, the MPC online solving optimization problem can be avoided.

The invention provides an aerial video image target tracking method and system based on broken trajectory repair, and the method comprises the steps: inputting an aerial video image into a sliding window slicing module frame by frame, and segmenting each frame of image into a plurality of image slices P1-Pn; respectively inputting the original images P0 and P1-Pn of each frame into a trained joint detection-tracking model, wherein the model comprises a target detection network and a Re-ID network integrated with a broken trajectory prediction and restoration algorithm; the target detection network carries out target detection on P0-Pn; and the Re-ID network performs target tracking according to the target detection result of each target of each frame: initializing the detection frame coordinate of each target of the first frame into a target trajectory, matching the detection frame coordinate of each target of the current frame with the existing target trajectory by combining the ID identity vector from the second frame, and if the matching succeeds, performing target tracking. And updating the detection frame coordinate of each target of the current frame to the target track, and if the matching is not successful, creating a new target track. According to the invention, target tracking can be carried out on aerial video images.