76 results about "Pedestrian behavior" patented technology

The invention provides a privacy protection method for multi-modal pedestrian behavior monitoring, and relates to the field of information security, and the method comprises the following steps: 1, carrying out the multi-modal pedestrian spatial-temporal feature fusion through a multi-modal pedestrian data fusion method based on spatial-temporal hybrid Mama; step 2, effectively protecting fusion features based on a differential privacy protection method of adaptive feature density perception; 3, a multi-modal pedestrian behavior pre-training method based on a double-flow dynamic Transform is adopted, and the pre-training precision of the model is improved; 4, a sparse gradient privacy protection method oriented to adaptive space-time correlation is adopted to ensure the safety of the model; and step 5, performing efficient monitoring based on a multi-modal pedestrian behavior monitoring method of a dynamic learning rate. According to the method, high-accuracy pedestrian behavior monitoring is realized, # imgabs0 # differential privacy is met based on the differential privacy property, and the safety and effectiveness of multi-mode pedestrian behavior monitoring are improved.

The invention discloses a building pedestrian simulation statistics and analysis method, and belongs to the technical field of simulation. According to the invention, a comprehensive pedestrian behavior evaluation and optimization scheme is provided through normal pedestrian circulation simulation and emergency evacuation simulation in combination with statistical analysis, data visualization and an optimization algorithm. The system manages pedestrian density through a KD tree, performs real-time data visualization by using ScottPlot, generates one-dimensional, two-dimensional and three-dimensional charts, and visually displays pedestrian distribution and flow conditions in a building. In an emergency evacuation scene, the system dynamically plans a path through a # imgabs0 # algorithm, and identifies and relieves a potential bottleneck area in combination with a heuristic function of congestion measurement. The simulation system can collect abundant statistical data including evacuation time, people flow distribution, congestion point analysis and the like, and provides a scientific basis for building safety design improvement and emergency plan making.

The invention discloses a pedestrian behavior simulation operation control method based on an unreal engine. Efficient and vivid pedestrian behavior simulation is realized through five core modules: a three-dimensional scene construction module converts a BIM model based on an unreal engine Datasmith plug-in and constructs an interactive dynamic scene; the pedestrian model construction module defines multi-dimensional pedestrian attributes and supports data-driven initial distribution configuration; the hierarchical behavior control module realizes refined regulation and control of pedestrian behaviors through macroscopic path planning, microcosmic behavior decision and emotion driving; the dynamic resource optimization module adopts level-of-detail control and multi-thread task allocation to improve the operation efficiency of the system; and the real-time interaction and data analysis module provides scene regulation and control, data visualization and simulation data recording and playback functions. According to the method, the authenticity, the dynamic response capability and the controllability of pedestrian simulation are remarkably improved, and reliable technical support can be provided for traffic planning, emergency drilling and building design evaluation.

The invention discloses a road traffic abnormal state monitoring system, belongs to the technical field of image recognition, and solves the problem that in the process that a bus stops at a stop, a ghost probe is most likely to appear at the position of a bus head, pedestrians may leave from the position of the bus head and cross a road, and the pedestrians are likely to go wrong in the process. The road monitoring cannot shoot the behavior that the pedestrian leaves from the position of the bus head and tries to cross the road. Comprising a license plate recognition sub-module, an equipment authority control module, a pedestrian behavior detection module, a decision analysis engine module, a data storage management module and a real-time early warning execution mechanism. According to the invention, through the multi-mode perception fusion technology, the identity of a bus is identified in real time, the motion trail of pedestrians is monitored, the collision risk is dynamically evaluated, an obstacle avoidance strategy is generated based on a vehicle dynamics model and multi-source data fusion, and encrypted data storage and hierarchical early warning response are implemented. And full-process active protection of the'ghost probe 'accident is realized when the bus enters the station.

The invention relates to the technical field of traffic planning and management optimization, and provides a regional hub and urban rail transit interchange connection optimization method, which comprises the following steps of S1, adopting different models for performing short-time arrival passenger flow prediction aiming at different travel modes to obtain respective interchange passenger flow; s2, modeling is conducted on the environment layout and pedestrian behaviors of an actual station through Anylogic software, a dynamic evaluation index system is constructed on the basis of the whole hub and the basic transfer facility state, passenger flow prediction data are input into simulation software, and transfer connection efficiency is obtained; and S3, constructing a maximum comprehensive benefit function from three aspects of a comfort index, an economic index and a facility traffic capacity balance degree, and optimizing transfer service facility configuration by adopting a genetic algorithm to optimize transfer connection efficiency. According to the optimization method provided by the invention, the transfer passenger flow volume can be accurately predicted, and the resource allocation is reasonably optimized, so that the transfer efficiency is improved, and the experience feeling of passengers is improved.

The embodiment of the invention provides a trajectory prediction method and device and a vehicle. The method comprises the steps that environment information around a vehicle, pedestrian information, intention probability and scene probability are acquired, and the intention probability is used for representing the probability that pedestrians execute each predefined pedestrian behavior mode in each predefined typical traffic scene; the scene probability is used for representing the probability that the traffic scene where the vehicle is located currently is the typical traffic scene; and performing trajectory prediction on the pedestrian based on the environment information, the pedestrian information, the intention probability, the scene probability and a pre-constructed space-time prediction network. According to the method, the scene probability and the intention probability are utilized to guide the space-time prediction network to learn a pedestrian motion mode under scene constraint and intention guidance, and the trajectory prediction accuracy can be improved.

The invention provides a forklift and pedestrian trajectory prediction and collision early warning method based on multi-camera fusion, and relates to the technical field of forklift and pedestrian trajectory prediction and collision early warning. The method comprises the following steps: respectively detecting the moving states of forklifts and pedestrians in a factory building, pedestrian behavior characteristics, driver driving data and forklift characteristics, predicting pedestrian walking paths according to pedestrian basic prediction data, the moving states and current working contents by establishing a pedestrian trajectory prediction model, and predicting the pedestrian walking paths by establishing a forklift trajectory prediction model. According to basic prediction data, moving states and current loading and unloading contents of a forklift, a forklift carrying route is planned, and collision early warning is carried out according to track prediction of pedestrians and the forklift by carrying out real-time driving safety scoring and forklift state safety scoring on a driver, judging the integrating degree of the forklift and setting the early warning time of the forklift.

The invention relates to a vehicle warning tone dynamic adjustment method and system based on multi-dimensional perception. According to the method, environment data are acquired, a multi-dimensional feature matrix is constructed, scene classification labels and sound field coverage parameters are analyzed based on the matrix, then an acoustic physical model is called to generate an environment self-adaptive acoustic parameter group, pedestrian behavior prediction and regional voice features are fused to implement multi-modal optimization on acoustic parameters to generate a prompt tone template, and the prompt tone is obtained. A directional sound field signal is synthesized through a beam forming technology, pedestrian response data is collected, finally, a parameter mapping rule and an acoustic physical model are continuously updated based on feedback information, a technical path of environmental perception, dynamic parameter adjustment, directional output and closed-loop optimization is formed, the self-adaptive capacity of the prompt tone system to complex traffic scenes is achieved, and the prompt tone system has a wide application prospect. While the warning effectiveness is guaranteed, the environmental interference is reduced, different regional cultural characteristics and pedestrian perception characteristics are considered, and the safety and environmental friendliness of human-vehicle interaction are integrally improved.

A method of generating pedestrian behavior prediction information includes setting image information from present to a certain time in the past as the target observation image, extract multiple visual input feature information and non-visual input feature information from the target observation image, grouping the multiple visual input feature information and non-visual input feature information, and inputting the grouped information to separated processing modules and generating pedestrian behavior prediction information by concatenating output results of the processing modules.

A control system of a vehicle detects a close-call vehicle-pedestrian encounter where the vehicle and a nearby object of concern almost collide and cause and accident, collects data for a previous period before the detected close-call vehicle-pedestrian encounter, the collected data including data captured by a set of perception sensors of the vehicle during the previous period, and transmits the collected data to a computing server configured to train a vehicle-pedestrian encounter model based on the collected data. The computing server then receives vehicle information indicative of a current state of the vehicle and executes the trained vehicle-pedestrian encounter model using the vehicle information to predict a future potential vehicle-pedestrian encounter and transmits encounter information indicative of the future potential vehicle-pedestrian encounter to the control system, which selectively generates an alert indicative of the future potential vehicle-pedestrian encounter for a driver of the vehicle.

The invention discloses a robot autonomous navigation method based on reinforcement learning in a complex dynamic environment, and relates to the technical fields of reinforcement learning, robot navigation and the like. The method aims at solving the problems that in multi-target reinforcement learning, conflicts exist between targets, so that navigation strategies are difficult to dynamically balance, and the decision-making flexibility is insufficient. The method comprises the following steps: firstly, creating a simulation environment, and constructing a Bayesian-RVO model for simulating pedestrian behaviors in the simulation environment; secondly, an equilibrium response principle in the game theory is introduced to design a reward function, and a game theory framework is constructed; then, designing a double-layer decision network based on a Stackelberg game structure; on the basis, a Critic network of a BNN structure is introduced; and finally, carrying out model training. Compared with the prior art, the method has the advantages that higher navigation success rate, lower collision rate and higher path reasonability are shown in the complex environment, the autonomous navigation performance of the robot in the complex environment is remarkably improved, and the method can be applied to the fields of agriculture, industry, service industry and the like.

The invention provides an intelligent driving obstacle avoidance method and device based on pedestrian attitude trajectory prediction, equipment and a storage medium, and belongs to the technical field of intelligent driving. Comprising the following steps: inputting an obtained four-channel RGB image, position information, speed, acceleration, direction, attitude and skeleton key points of a target pedestrian within a preset time length into a preset attitude trajectory prediction model, and predicting coordinates of the skeleton key points of the target pedestrian; and determining an intelligent obstacle avoidance strategy according to the coordinates of the skeleton key points and the driving path data of the target vehicle. According to the invention, the prediction accuracy of the intelligent driving system on pedestrian behaviors can be improved, and the safety of obstacle avoidance decision making is enhanced.

The invention relates to an ultra-large space pedestrian flow simulation system, and relates to the technical field of pedestrian flow management technologies, and the system comprises an autonomous decision behavior simulation module which simulates the behavior decision of pedestrians in a complex space, and the behavior decision comprises path selection, speed adjustment and avoidance behaviors; the environment sensing module is used for identifying obstacles, channel layout, exit positions and other environment characteristics in the space; the time period flow simulation module simulates pedestrian flow density and distribution in different time periods, and analyzes and optimizes configuration of various service facilities; and the visual display module is used for displaying a simulation result through a thermodynamic diagram, a traffic matrix and evacuation path tracking. According to the method, factors such as pedestrian behaviors, environmental perception and density are comprehensively considered, the optimal path is generated, pace is dynamically adjusted, congestion and collision are avoided, and meanwhile, the system senses environmental changes in real time and predicts the future flow trend through sensor data fusion and machine learning technologies.

The invention provides a behavior recognition method based on a visual angle of a vehicle-mounted camera and deep learning attitude estimation, and the method comprises the steps: collecting an image frame of a front road through a vehicle-mounted foresight camera, and obtaining the speed of a vehicle through an OBD sensor; performing attitude estimation on the pedestrian in the image frame through an OpenPose model to extract pedestrian attitude key point coordinate information and generate a pedestrian bounding box; providing a pedestrian ID for each pedestrian by adopting a pedestrian bounding box generated by combining a tracking algorithm with attitude estimation; and inputting the pedestrian posture key point coordinate information, the self-vehicle speed and the pedestrian ID into a trained behavior recognition model for behavior classification, and outputting a pedestrian behavior recognition result. According to the method, the accuracy and robustness of pedestrian behavior recognition can be improved from multiple dimensions, the method is particularly suitable for application scenes where dynamic behaviors of pedestrians need to be analyzed, the behaviors of the pedestrians can be quickly and accurately recognized, and the intelligent level of vehicles and the safety of the pedestrians are enhanced.

The invention discloses a pedestrian circulation and personnel evacuation evaluation simulation system in a public building, and relates to the field of personnel evacuation simulation, comprising personnel circulation simulation behaviors and personnel evacuation evaluation simulation behaviors in the public building, and pedestrian circulation simulation analyzes behavior modes and dynamic changes of pedestrians in different scenes. The personnel evacuation evaluation simulation adopts a multi-agent modeling technology to research and develop a bottom-layer pedestrian simulation engine, pedestrian behavior planning in a bottom-layer pedestrian simulation model is combined with a behavior tree and a finite-state machine, and a flow chart connection type behavior node editor is adopted as user behavior input; the bottom layer pedestrian simulation engine realizes dynamic simulation based on pedestrian dynamics, and an ORCA collision algorithm is adopted in the pedestrian dynamics to ensure that an intelligent agent plans a path in an optimal mode; in pedestrian dynamics, a steering force model SteeringBehavior is adopted to simulate acceleration and steering behaviors. By adopting the method, the performance of personnel evacuation simulation and the expansion capability are improved.

The invention discloses a pedestrian bowing behavior detection method and system for an intelligent garbage booth, and relates to the technical field of intelligent Internet of Things, and the method comprises the steps: employing YOLOv8-Pose to detect key point coordinates and bounding boxes, generating a multi-scale optical flow field through an RAFT algorithm, constructing a sampling set, obtaining an affine transformation matrix and an optical flow residual error, marking an abnormal point, and carrying out the detection of a pedestrian bowing behavior. A replacement optical flow is generated by optimizing an objective function, a spatio-temporal feature tensor is constructed and input into a Transform encoder, smooth key points are output, and updated key point coordinates are generated by screening and complementing; according to the method, through YOLOv8-Pose attitude estimation and RAFT multi-scale optical flow fusion, the precision and robustness of bowing behavior detection are improved, and through Transform space-time coding and HMM state sequence optimization fusion, the real-time performance and reliability of intelligent garbage booth pedestrian behavior detection are improved.

The invention relates to the technical field of image processing, and discloses an idle scooter detection method and system. The method comprises the following steps: acquiring a multi-source video through a city monitoring camera, preprocessing the multi-source video, identifying a scooter by using improved YOLOv8, constructing a self-adaptive region of interest through polar coordinate mapping, extracting pedestrian behavior characteristics by using a double-flow network, and calculating interaction strength in combination with space intersection, time continuity and posture identification. And finally, judging the idle state of the scooter based on the hierarchical decision-making tree and scoring. According to the method, the idle state of the shared scooter in the urban environment can be accurately identified and judged in real time, and high-precision use state evaluation is provided.

The present invention provides a posture recognition method, system, device and medium, comprising: performing self-supervised learning on unlabeled posture sequence images to generate a pre-trained model; inputting labeled posture sequence images into the pre-trained model for iterative training, and adjusting the parameters of the pre-trained model during the iterative training process to generate a posture recognition model for identifying the posture of a target object. The present invention can improve the model's ability to represent human posture and the performance of pedestrian behavior recognition tasks by using a large amount of unlabeled data when there is less labeled data. Moreover, the self-supervised comparative learning method proposed in the present invention effectively utilizes the person ID information and time-series frame information in the human image data, especially by using the same person ID and adjacent time-series images as negative example pairs, which can directly ignore the portrait information within the same person ID and focus on the changes in human posture.

The invention relates to an unmanned aerial vehicle detection-based method and system for detecting pedestrians who do not give way to a vehicle, and the method comprises the steps: carrying out the image collection through an image collection device carried by an unmanned aerial vehicle, and obtaining a traffic image sequence; performing multi-target detection on all moving targets in the frame image through a target detection model to obtain a target detection frame and corresponding confidence; performing trajectory tracking processing on the target detection frames and the confidence coefficients of all the frame images through a target tracking algorithm to obtain a tracking trajectory of the moving target; obtaining a zebra crossing control area, calculating a pedestrian warning point and a vehicle warning point based on the tracking trajectory when the zebra crossing control area in the frame image is detected to have the pedestrian trajectory and the vehicle trajectory, and calculating a vehicle dangerous area according to the vehicle warning point; and according to the tracking trajectory, the pedestrian warning point and the vehicle dangerous area, analyzing a position relationship between the pedestrian and the vehicle, and according to the position relationship and a preset non-comity pedestrian judgment logic, determining whether there is a behavior that the vehicle does not comity the pedestrian.

The application provides a two-phase signal intersection pedestrian nested phase optimization model and a judgment method for inserting a pedestrian nested phase, solves the conflict between existing intersection pedestrians and turning vehicles, and specifically designs a control system for the information interaction of the intersection street-crossing pedestrians and the passing vehicles and the signal lights.In the signal light pedestrian nested phase setting condition, a total cost (including two parts of the conflict number and the delay) of the traffic participants is used as an optimization target, a pedestrian nested phase total cost model is provided, the motor vehicle avoidance pedestrian behavior in the human-vehicle interaction behavior is considered to be different in different cities and different intersections, and therefore, the suggestion value of whether to set the pedestrian nested phase is given under different traffic flow environments, the human-vehicle can pass through the intersection under the condition that the traffic operation cost is minimum, the efficiency and safety of the human-vehicle passing through the intersection are ensured, and the efficiency and safety of the human-vehicle passing through the intersection are ensured.

The invention discloses a pedestrian trajectory prediction method based on lightweight interactive perception GAN, and the method comprises the following steps: S101, carrying out the collection and preprocessing of pedestrian historical trajectory data collected based on a camera in a closed scene, S102, carrying out the feature extraction of a pedestrian and an interaction object based on a distance weight, carrying out the quantitative expression of an interaction relation, and carrying out the prediction of the pedestrian trajectory. The method comprises the following steps: step S103, constructing and training a lightweight GAN model based on a GRU and a double-constraint discriminator to obtain a trained lightweight GAN model, step S104, obtaining and outputting final predicted trajectory data based on trajectory prediction and effectiveness judgment of the trained GAN model, and step S105, carrying out differentiated visual output on historical and predicted trajectories. The method can be widely applied to pedestrian behavior analysis scenes of closed scenes such as shopping mall passenger flow guidance, community security and protection monitoring and supermarket moving line planning, and provides data support for scene operation management and safety prevention and control.

The present invention discloses a pedestrian circulation and evacuation assessment simulation system in public buildings, relating to the field of evacuation simulation. The system includes simulating pedestrian circulation and evacuation assessment in public buildings. The pedestrian circulation simulation analyzes pedestrian behavior patterns and dynamic changes in different scenarios. The evacuation assessment simulation utilizes multi-agent modeling technology to develop an underlying pedestrian simulation engine. Within the underlying pedestrian simulation model, pedestrian behavior planning combines behavior trees and finite state machines, using a flow chart-style behavior node editor as user behavior input. The underlying pedestrian simulation engine implements dynamic simulation based on pedestrian dynamics. The ORCA collision algorithm is used in pedestrian dynamics to ensure that agents plan paths in an optimal manner. A steering force model, SteeringBehavior, is used in pedestrian dynamics to simulate acceleration and steering behavior. The present invention utilizes the aforementioned methods to improve the performance and expandability of evacuation simulation.

The application discloses a light neural network model edge end cross identification vehicle courtesy pedestrian method and system, belongs to the technical field of identifying vehicle courtesy pedestrians, and comprises the following steps: S1: a vehicle-mounted road surface video collector of a bus video acquisition system collects three groups of picture data of no zebra crossing picture D1, no pedestrian on the zebra crossing D2 and pedestrian on the zebra crossing D3 in a driving process; S2: the collected picture data is subjected to image preprocessing; S3: a zebra crossing identification model is trained by using negative samples (D2+D3) and positive samples D1, and a pedestrian identification model on the zebra crossing is trained by using positive samples D3 and negative samples D2; and S4: when a camera in the vehicle does not detect a zebra crossing on the road, only the zebra crossing identification model is kept listening, and when the zebra crossing on the road is detected; listening data is selectively collected to consider a multidimensional interaction state of a vehicle and a target traffic light area road condition, and a vehicle courtesy pedestrian behavior is quantified in a scoring mode.

Provided are methods for simulated smart pedestrians, The method includes obtaining attributes of at least one pedestrian dynamics model. Simulated sensor data associated with the environment is generated. Operation of an autonomous system in the environment is simulated based on the simulated sensor data, wherein vehicle-pedestrian interactions are modeled by the at least one pedestrian dynamics model as external forces in the environment impacting behavior of the respective pedestrian.

The application provides a commercial space value evaluation and optimization method and system based on pedestrian simulation, relates to the field of simulation, and comprises the following steps: constructing a three-dimensional mall model, including building structure, stall and vertical traffic data; setting a dynamic time engine according to the obtained external data, the dynamic time engine converts the obtained external data and is used to drive the flow of people flow when driving the pedestrian simulation; based on the constructed three-dimensional mall model and the dynamic time engine, a pedestrian simulation model based on commercial simulation is constructed, a reward function is set, a PPO algorithm is used for automatic optimization, and a comprehensive optimal three-dimensional mall model is obtained. The application adopts the above-mentioned commercial space value evaluation and optimization method and system based on pedestrian simulation, realizes the cross-scale mapping from micro pedestrian behavior to macro commercial value, and provides an interpretable, intervenable digital twin platform for commercial space optimization.

The application discloses a pedestrian interaction behavior recognition method in a complex scene, comprising: collecting pedestrian behavior videos in a preset period in a target scene, obtaining continuous trajectory data of each pedestrian and converting the continuous trajectory data into original trajectory data in a real world coordinate system, and obtaining target scene pedestrian motion trajectory data to be recognized through preprocessing; constructing initial recognition rules including initial preset rules and initial to-be-optimized rules; determining a dynamic optimization strategy of the initial to-be-optimized rules according to the target scene, optimizing the initial to-be-optimized rules, obtaining target dynamic recognition rules, and recognizing the target scene pedestrian motion trajectory data according to the rules to determine pedestrian interaction behaviors. The method can realize accurate recognition of pedestrian interaction behaviors, provides a complete pedestrian interaction behavior recognition method from data collection and processing to final behavior determination, and can provide efficient and accurate data support for urban traffic management, public security monitoring and social behavior research.

The application provides a forklift and pedestrian trajectory prediction and collision warning method based on multi-camera fusion, and relates to the technical field of forklift and pedestrian trajectory prediction and collision warning. The application acquires panoramic and regional images in a factory building, establishes panoramic and regional detection models, detects the moving state of forklifts and pedestrians, pedestrian behavior characteristics, driver driving data and forklift characteristics in the factory building respectively, establishes a pedestrian trajectory prediction model, predicts the pedestrian walking path according to the pedestrian basic prediction data, moving state and current work content, establishes a forklift trajectory prediction model, plans the forklift carrying route according to the forklift basic prediction data, moving state and current loading and unloading content, judges the fitting degree of the forklift through real-time driving safety scoring of the driver and forklift state safety scoring, sets the warning time of the forklift, and performs collision warning according to the trajectory prediction of the pedestrian and the forklift.

This invention relates to the field of safe traffic technology and proposes a pedestrian crossing safety signal and control system. The system comprises a data acquisition module for acquiring pedestrian status, identifying special groups of people, and obtaining vehicle type, speed, and weight data; a behavior analysis module, connected to the data acquisition module via data transmission technology, for predicting pedestrian behavior trends and vehicle passage; and a data analysis module, connected to the behavior analysis module via data transmission technology, for analyzing predicted data to determine whether pedestrians are allowed to cross the zebra crossing and estimating the duration of the crossing. By capturing changes in pedestrian walking status in real time and considering factors such as the possible speed recovery situation (PSRS) and the impossible speed recovery situation (ISRS) in the event of unexpected situations while crossing the zebra crossing, the system effectively prevents pedestrians from being stranded in the road due to improperly set signal light timing, significantly improving the safety and smoothness of pedestrian crossings.

The invention relates to the technical field of campus safety monitoring, and discloses a campus pedestrian abnormal behavior identification method based on data knowledge fusion. The method comprises the following steps: receiving real-time pedestrian data of a campus monitoring system, comparing the real-time pedestrian data with historical pedestrian behavior data, and extracting behavior characteristic components; using a behavior pattern recognition algorithm to recognize an abnormal behavior pattern from the feature components; performing association analysis on the abnormal modes through a feature importance evaluation algorithm to obtain key behavior factors; taking the key factors as input, constructing a reference behavior curve in combination with a normal behavior model, evaluating pedestrian behaviors according to the degree of deviation between the reference behavior curve and an actual behavior curve, and determining a potential abnormal risk level; and in combination with the abnormal mode, the key factor and the risk level, adopting a time sequence prediction model to predict the pedestrian with abnormal behavior in a preset time, generating an abnormal report, and assisting campus safety management.

The application discloses an intelligent agent multi-target navigation method based on deep reinforcement learning in a dynamic environment, and aims at the problems of separation of pedestrian behavior prediction and decision, insufficient modeling of behavior difference and insufficient multi-target trade-off in the prior art, and constructs a decision framework integrating pedestrian impatience prediction and Nash game. First, a simulation environment containing intelligent agents and pedestrians is established, and state and action spaces are designed; second, a multi-target reward function of safety, sociality and efficiency is constructed, and a meeting angle is introduced to depict the interaction relationship; further, a pedestrian impatience evolution model is established to realize dynamic prediction of the behavior trend of pedestrians; finally, the prediction result is embedded into a Nash equilibrium game model to guide the optimal decision; a safety penalty and a cost Critic network are introduced on the basis of a Soft Actor-Critic algorithm to realize parameter collaborative optimization.