Implementation method of unmanned intelligent public transportation system

Abstract

The invention discloses a realization method of an unmanned intelligent public transportation system, and relates to the technical field of automatic driving. Comprising the steps of 1, carrying out architecture layering design on the unmanned intelligent public transportation system, the architecture layering comprising a sensing layer, a decision-making layer, an execution layer, a communication layer, a cloud platform layer, an energy layer, a user interaction layer and a safety and redundancy layer, and 2, designing a sensor module, a data fusion analysis module and a fault tolerance module on the sensing layer, 3, designing a driving strategy planning module at a decision-making layer, 4, designing an execution module at an execution layer, 6, designing a cloud platform management module at a cloud platform layer, 7, designing an energy management module and a motor control module at an energy management layer, 8, designing an interaction module at a user interaction layer, and 9, designing a safety redundancy module at a safety and redundancy layer. And the security redundancy module monitors hardware faults, software exceptions or external attacks, deploys redundancy for hardware, configures double hardware, performs network data transmission by using an encryption technology, and performs data backup by using distributed storage.

Description

Technical Field

[0001] The present invention discloses a method for implementing an unmanned intelligent bus system, which relates to the technical field of autonomous driving. Background Art

[0002] An unmanned bus refers to a bus that automatically drives without a driver. Currently, it mainly uses a sensor system to autonomously sense the environment, identify roads, and avoid obstacles. However, in the face of complex environments or complex usage scenarios, the existing unmanned bus system is still unable to cope and cannot guide the bus to drive without human control. Summary of the Invention

[0003] In view of the problems of the prior art, the present invention provides a method for implementing an unmanned intelligent bus system, which comprehensively considers aspects such as autonomous driving technology, Internet of Things (IoT), artificial intelligence (AI), communication network, energy management, passenger experience, and safety redundancy to improve the unmanned intelligent bus system and achieve true driverless operation.

[0004] The specific solution proposed by the present invention is as follows:

[0005] The present invention provides a method for implementing an unmanned intelligent bus system, including:

[0006] Step 1: Conduct an architecture hierarchical design for the unmanned intelligent bus system. The architecture hierarchy includes a perception layer, a decision-making layer, an execution layer, a communication layer, a cloud platform layer, an energy layer, a user interaction layer, and a safety and redundancy layer.

[0007] Step 2: Design a sensor module, a data fusion and analysis module, and a fault tolerance module in the perception layer. The sensor module collects real-time data on the internal and external environments of the vehicle, providing information input for subsequent decision-making and execution. The data fusion and analysis module fuses the data collected by the sensor module and uses a deep learning convolutional neural network (CNN) to identify traffic signs and pedestrian intentions, and uses a target detection algorithm to detect dynamic obstacles in real time. At the same time, the fault tolerance module cross-verifies the data of multiple sensor modules. When a certain sensor module fails, it indicates that the data collected by other sensor modules is used for data replacement.

[0008] Step 3: Design a driving strategy planning module in the decision-making layer. The driving strategy planning module generates path planning and behavior decisions based on the data provided by the perception layer, and generates driving instructions according to the path planning and behavior decisions. The path planning includes global path planning and local path planning, and the behavior decision includes a rule engine. Preset traffic rules are used as decision boundaries, and a finite state machine (FSM) is used to manage the vehicle state.

[0009] Step 4: Design an execution module in the execution layer. The execution module converts the driving instructions generated by the decision-making layer into actual actions of the vehicle.

[0010] Step 5: Design a communication module at the communication layer. The communication module is responsible for data transmission between vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and vehicle-to-cloud (V2C).

[0011] Step 6: Design a cloud platform management module at the cloud platform layer. The cloud platform management module uses a distributed architecture for global scheduling, data analysis, and remote monitoring to achieve coordination among vehicles, passengers, and transportation infrastructure. It provides scheduling services and monitoring services using microservices and performs local decision-making and data processing at edge computing nodes to reduce reliance on the cloud.

[0012] Step 7: Design an energy management module and a motor control module at the energy management layer. The energy management module monitors the battery and real-time battery status, and uses an active balancing method to extend the battery life. The motor control module monitors the real-time operating status of the motor and uses a vector control algorithm (FOC) to improve motor efficiency.

[0013] Step 8: Design an interaction module at the user interaction layer. The interaction module uses a face recognition model for face recognition boarding, provides QR code / NFC scanning for boarding, and provides a multi-language interface. It supports multi-language interaction using the automatic speech recognition (ASR) method.

[0014] Step 9: Design a safety redundancy module at the safety and redundancy layer. The safety redundancy module monitors hardware failures, software anomalies, or external attacks, deploys redundant hardware, configures dual hardware, uses encryption technology for network data transmission, and uses distributed storage for data backup.

[0015] Furthermore, in step 2 of the method for implementing an unmanned intelligent bus system, the sensor module is configured with a lidar to generate a 3D point cloud map for obstacle detection; a multi-camera is configured through the sensor module to cover the surrounding of the vehicle for traffic sign recognition, lane line recognition, pedestrian recognition, and emergency situation recognition; an ultrasonic sensor is configured through the sensor module for close-range parking and obstacle avoidance in low-speed scenarios.

[0016] The data fusion and analysis module selects the Kalman filter and particle filter algorithms to fuse the data collected by the lidar and the multi-camera to generate a unified environmental model.

[0017] Furthermore, in step 3 of the method for implementing an unmanned intelligent bus system, the driving strategy planning module generates a path planning based on the data provided by the perception layer, including: using the A* algorithm and Dijkstra algorithm to generate a global path planning, which is the optimal path from the starting point to the ending point, dynamically adjusting the path by combining the map and real-time road conditions, and using the dynamic window approach (DWA) or model predictive control (MPC) to generate a smooth local path.

[0018] Behavioral decision-making includes collaborative decision-making. Based on the collaborative decision-making, the traffic light status and the position information of other vehicles are obtained through C-V2X to control the platooning of multiple vehicles, reducing wind resistance and energy consumption.

[0019] Furthermore, in step 4 of the implementation method of the unmanned intelligent bus system described above, the execution module converts the driving instructions generated by the decision-making layer into actual actions of the vehicle, including:

[0020] The execution module converts the steer-by-wire of the vehicle, adjusts the steering wheel angle according to the path planning instruction, configures a servo motor to drive the steering mechanism to steer, and at the same time uses the drive-by-wire to adjust the vehicle speed according to the road conditions and traffic flow, and configures a high-performance motor controller to adjust the speed.

[0021] The execution module configures the motor control to use the vector control algorithm FOC to control the motor, and dynamically adjusts the motor output power according to the road conditions; and recovers kinetic energy during coasting or braking and stores it in the battery.

[0022] Furthermore, in step 5 of the implementation method of the unmanned intelligent bus system described above, the communication module is responsible for the data transmission of vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and vehicle-to-cloud (V2C), including:

[0023] The communication protocol and standard use V2X communication, use the Cellular Vehicle-to-Everything (C-V2X) protocol, and are compatible with the Dedicated Short Range Communications (DSRC) protocol for interoperability with existing infrastructure;

[0024] The communication module uses the Controller Area Network (CAN) bus and Ethernet high-speed communication protocols to achieve efficient data interaction among the subsystems in the vehicle, supports the Time-Sensitive Networking (TSN) to ensure the real-time transmission of critical data.

[0025] Furthermore, in step 6 of the implementation method of the unmanned intelligent bus system described above, in the cloud platform layer, the Distributed File System (HDFS) is used to store massive data, and the big data frameworks Hadoop or Spark are used for data cleaning, analysis, and mining to generate operation reports to provide support for decision-making;

[0026] The global scheduling intelligent scheduling algorithm is used to optimize bus routes and schedules, and combined with real-time road conditions and passenger flow data, the scheduling strategy is dynamically adjusted; the remote monitoring uses visualization tools such as Grafana or Kibana to monitor the vehicle status and operation in real time and set up a warning mechanism.

[0027] Furthermore, in step 7 of the implementation method of the unmanned intelligent bus system described above, the energy management module also dynamically adjusts the motor output power according to the road conditions and passenger capacity, uses a prediction algorithm to optimize the energy consumption strategy, intelligently schedules charging according to the battery status and trip plan, and uses a linear programming algorithm to minimize the charging cost and shorten the charging time.

[0028] Furthermore, in step 8 of the implementation method of the unmanned intelligent bus system, the interaction module also uses a message queue to push bus information in real time, uses a recommendation algorithm to provide personalized services, and controls barrier-free facilities to serve special groups. The interaction module pushes information through touch screens, cameras, and microphone devices.

[0029] Furthermore, in step 9 of the implementation method of the unmanned intelligent bus system: The security redundancy module uses encryption technology for network data transmission, uses the TLS / SSL protocol to encrypt the communication link to prevent data leakage, and uses digital certificates to authenticate both communication parties to prevent man-in-the-middle attacks;

[0030] Deploy a firewall and an intrusion detection system IDS to monitor network traffic in real time, and use blockchain technology to record communication logs to prevent data tampering.

[0031] The present invention also provides an unmanned intelligent bus system, including:

[0032] The sensor module, data fusion and analysis module, and fault tolerance module in the perception layer. The sensor module collects vehicle internal and external environment data in real time, providing information input for subsequent decision-making and execution. The data fusion and analysis module fuses the data collected by the sensor module, and uses a deep learning convolutional neural network CNN to identify traffic signs and pedestrian intentions, and uses a target detection algorithm to detect dynamic obstacles in real time. At the same time, the fault tolerance module cross-verifies the data of multiple sensor modules. When a certain sensor module fails, it instructs the data collected by other sensor modules to be used for data replacement.

[0033] The driving strategy planning module in the decision-making layer. The driving strategy planning module generates path planning and behavior decisions based on the data provided by the perception layer, and generates driving instructions according to the path planning and behavior decisions. The path planning includes global path planning and local path planning. The behavior decision includes a rule engine, with preset traffic rules as decision boundaries, and uses a finite state machine FSM to manage the vehicle state.

[0034] The execution module in the execution layer. The execution module converts the driving instructions generated by the decision-making layer into actual actions of the vehicle.

[0035] The communication module in the communication layer. The communication module is responsible for data transmission between vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and vehicle-to-cloud (V2C).

[0036] The cloud platform management module in the cloud platform layer. The cloud platform management module uses a distributed architecture for global scheduling, data analysis, and remote monitoring, realizing the coordination of vehicles, passengers, and traffic infrastructure, providing scheduling services and monitoring services using microservices, and performing local decision-making and data processing at edge computing nodes to reduce dependence on the cloud.

[0037] The energy management module and motor control module of the energy management layer. The energy management module monitors the battery, real-time monitors the battery status, and uses the active balancing method to extend the battery life. The motor control module real-time monitors the motor operation status and uses the vector control algorithm FOC to improve the motor efficiency.

[0038] The interaction module of the user interaction layer. The interaction module uses the face recognition model for face recognition to get on the bus, provides QR code / NFC scanning to get on the bus, and provides a multi-language interface, and uses the speech recognition ASR method to support multi-language interaction.

[0039] The safety and redundancy module of the safety and redundancy layer. The safety and redundancy module monitors hardware failures, software anomalies or external attacks, deploys redundant hardware, configures dual hardware, and uses encryption technology for network data transmission and uses distributed storage for data backup.

[0040] The advantages of the present invention are:

[0041] It can improve the quality of public transportation services, reduce operating costs, reduce carbon emissions, and also promote the progress of fields such as autonomous driving, artificial intelligence, communication technology, and big data. At the same time, the driverless bus system will also promote policy innovation and the improvement of traffic management level, providing important support for the construction of smart cities. In the future, with the continuous progress of technology and the in-depth application, the driverless bus system will become an important part of urban transportation, providing a more convenient, efficient, and environmentally friendly travel mode for people. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0043] Figure 1 It is a schematic diagram of the system deployment architecture of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0044] The following further illustrates the present invention in conjunction with the drawings and specific embodiments, so that those skilled in the art can better understand the present invention and be able to implement it, but the embodiments given are not intended to limit the present invention.

[0045] The present invention provides a method for implementing a driverless intelligent bus system, including:

[0046] Step 1: Conduct an architecture hierarchical design for the driverless intelligent bus system. The architecture hierarchy includes a perception layer, a decision layer, an execution layer, a communication layer, a cloud platform layer, an energy layer, a user interaction layer, and a safety and redundancy layer.

[0047] Step 2: Design a sensor module, a data fusion and analysis module, and a fault tolerance module in the perception layer. The sensor module collects real-time data on the internal and external environment of the vehicle, providing information input for subsequent decision-making and execution. The data fusion and analysis module fuses the data collected by the sensor module and uses a deep learning convolutional neural network CNN to identify traffic signs and pedestrian intentions, and uses object detection algorithms to detect dynamic obstacles in real time. At the same time, the fault tolerance module cross-verifies the data of multiple sensor modules. When a certain sensor module fails, it indicates that the data collected by other sensor modules is used for data replacement.

[0048] Among them, the sensor module is configured with lidar to generate a high-precision 3D point cloud map to detect obstacles. One or two 360° rotating LiDARs are installed on the roof, and short-range LiDARs are installed on the front and rear bumpers; the sensor module is configured with multi-camera, such as front view, side view, and rear view to cover the surrounding of the vehicle to identify traffic signs, lane lines, pedestrians and emergencies; long-range radars are installed on the front and rear bumpers, and short-range radars are installed on the sides of the vehicle body for all-weather ranging and speed detection; multiple ultrasonic sensors are installed around the vehicle body, and the sensor module is configured with ultrasonic sensors for close-range parking and obstacle avoidance in low-speed scenarios; a high-precision GNSS receiver is configured and integrated with an inertial navigation system IMU module to provide vehicle position, speed, and attitude information. The data fusion module fuses multi-sensor data and uses algorithms such as Kalman filtering and particle filtering to fuse LiDAR, camera, and radar data to generate a unified environment model, such as the position, speed, and type of obstacles; a deep learning convolutional neural network CNN is used to identify traffic signs and pedestrian intentions, and object detection algorithms such as YOLO and Faster R-CNN are used to detect dynamic obstacles in real time. The fault tolerance module includes multi-sensor cross-verification. When a certain sensor fails, such as the camera being blocked, other sensors can provide alternative data.

[0049] Hardware redundancy configuration such as LiDAR and radar. When the main sensor fails, the backup sensor immediately takes over; the software fault tolerance perception algorithm has a built-in anomaly detection mechanism to filter out noise data, such as radar false alarms. The sensor data is transmitted using Ethernet or CAN bus to ensure low latency. At the same time, the perception algorithm runs on an in-vehicle computing unit (NVIDIA Drive AGX) to reduce the dependence on the cloud.

[0050] Step 3: Design a driving strategy planning module in the decision-making layer. The driving strategy planning module generates path planning and behavior decisions based on the data provided by the perception layer, and generates driving instructions according to the path planning and behavior decisions. The path planning includes global path planning and local path planning. The behavior decision includes a rule engine, preset traffic rules as decision boundaries, and uses a finite state machine FSM to manage the vehicle state.

[0051] Among them, path planning includes global path planning, which uses the A* algorithm and Dijkstra algorithm to generate the optimal path from the starting point to the ending point, and dynamically adjusts the path by combining high-precision maps and real-time traffic conditions; local path planning uses the Dynamic Window Approach (DWA) or Model Predictive Control (MPC) to generate a smooth local path while avoiding dynamic obstacles such as pedestrians and vehicles.

[0052] Behavior decision-making includes a rule engine that presets traffic rules such as stopping at red lights and yielding to pedestrians as decision boundaries, and uses a Finite State Machine (FSM) to manage vehicle states such as following and lane changing; the AI algorithm uses Reinforcement Learning (RL) to train the vehicle's decision-making ability in complex scenarios and uses deep learning models such as Long Short-Term Memory (LSTM) to predict the intentions of pedestrians and other vehicles. Cooperative decision-making includes Vehicle-to-Everything (V2X) communication, which obtains information such as traffic light status and the positions of other vehicles through Cellular-V2X (C-V2X), supports platooning of multiple vehicles to reduce wind resistance and energy consumption; the cooperative driving algorithm uses the Alternating Direction Method of Multipliers (ADMM), a distributed optimization algorithm, to achieve cooperative path planning for multiple vehicles. Emergency decision-making sets up a rapid response mechanism and uses rule-based emergency strategies such as emergency braking to handle emergencies, and combines data from the perception layer such as the sudden appearance of pedestrians to generate emergency instructions. The hardware and software platform includes a computing platform that uses a high-performance in-vehicle computing unit, NVIDIA Drive AGX, to run decision-making algorithms, supports multi-core parallel computing, and meets real-time requirements; the software architecture uses a modular design for easy algorithm iteration and upgrade, and supports Over-the-Air (OTA) software updates to update decision-making algorithms.

[0053] Step 4: Design an execution module in the execution layer. The execution module converts the driving instructions generated by the decision-making layer into actual actions of the vehicle.

[0054] The execution module converts the steer-by-wire of the vehicle, adjusts the steering wheel angle according to the path planning instructions, configures a high-precision motor-driven steering mechanism, and eliminates mechanical connections; the brake-by-wire achieves smooth deceleration or emergency braking according to the behavior decision-making instructions. Dynamically adjust the motor output power according to road conditions such as climbing and congestion; energy recovery recovers kinetic energy during coasting or braking and stores it in the battery, and uses intelligent algorithms to optimize the energy recovery efficiency. Additionally, configure dual-redundant hardware, such as having two sets of steering motors and brake hydraulic pumps. When the main controller fails, the backup system immediately takes over. Monitor the status of actuators such as temperature and pressure in real time and give warnings, and at the same time use Fault Tree Analysis (FTA) to improve the fault tolerance of the system. Maintain state feedback, configure a steering angle sensor to monitor the steering wheel angle in real time, a brake pressure sensor to monitor the brake system pressure in real time, and a motor speed sensor to monitor the motor speed in real time, and adjust the actuator actions in real time according to the sensor feedback to improve the control accuracy.

[0055] Step 5: Design a communication module in the communication layer. The communication module is responsible for data transmission between vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and vehicle-to-cloud (V2C).

[0056] Among them, the communication protocol and standard are for V2X communication. The Cellular Vehicle-to-Everything (C-V2X) protocol is used, which supports low-latency and high-reliability communication, is compatible with the Dedicated Short Range Communications (DSRC) protocol, and ensures interoperability with existing infrastructure; in-vehicle communication uses high-speed communication protocols such as Controller Area Network (CAN) bus and Ethernet to achieve efficient data interaction among various in-vehicle subsystems, supports Time-Sensitive Networking (TSN), and ensures real-time transmission of critical data. The network architecture uses a 5G network, deploys 5G base stations, supports high-bandwidth and low-latency communication between vehicles and the cloud, and uses network slicing technology to allocate dedicated network resources for autonomous driving. In terms of data security, encryption and authentication use the Transport Layer Security / Secure Sockets Layer (TLS / SSL) protocol to encrypt the communication link to prevent data leakage, and use digital certificates to authenticate both communication parties to prevent man-in-the-middle attacks; for intrusion detection, firewalls and intrusion detection systems (IDS) are deployed to monitor network traffic in real time, and blockchain technology is used to record communication logs to prevent data tampering. The on-vehicle communication unit of the hardware and software platform uses a high-performance communication module (5G module), which supports multi-band and multi-protocol communication, and is configured with a dual communication module (5G + DSRC) to achieve redundant design; the cloud platform deploys a distributed cloud platform, which supports massive data storage and processing, uses a microservices architecture, and supports high-concurrency access and rapid iteration.

[0057] Step 6: Design a cloud platform management module at the cloud platform layer. The cloud platform management module uses a distributed architecture for global scheduling, data analysis, and remote monitoring to achieve coordination among vehicles, passengers, and traffic infrastructure, provides scheduling services and monitoring services using microservices, and performs local decision-making and data processing at edge computing nodes to reduce dependence on the cloud.

[0058] Use the distributed file system HDFS to store massive amounts of data, supporting multiple data formats; for big data processing, use big data frameworks such as Hadoop and Spark for data cleaning, analysis, and mining to generate operation reports and provide support for decision-making. Global scheduling uses intelligent scheduling algorithms to optimize bus routes and schedules, and combines real-time traffic conditions and passenger flow data to dynamically adjust scheduling strategies; collaborative scheduling obtains information such as traffic signal status and the positions of other vehicles through V2X communication, optimizes traffic efficiency, supports multi-vehicle platooning, and reduces wind resistance and energy consumption. Remote monitoring uses visualization tools (Grafana, Kibana) to monitor vehicle status and operation in real time, and sets up early warning mechanisms, such as issuing an alarm when the battery level is lower than 20%; remote intervention supports remotely taking over abnormal vehicles and provides fault diagnosis and repair suggestions. OTA upgrade uses a secure OTA mechanism, such as digital signatures, to update in-vehicle software and algorithms, supports incremental updates, and reduces data transmission volume. At the same time, it provides passenger services, including real-time information push, using a message queue (Kafka) to push bus information in real time, such as arrival time and transfer suggestions, and supports multi-platform interaction, such as mobile apps and electronic bus stops; and personalized services, using a recommendation algorithm (collaborative filtering) to provide personalized services (such as reserved seats), and supports multi-language and multi-platform interaction to facilitate passenger use. Safety and privacy include data encryption, using encryption technologies (AES, RSA) to protect the security of data transmission and storage, supporting data desensitization to protect passenger privacy; access control uses identity authentication and permission management mechanisms to ensure that only authorized users can access data, and monitors network traffic in real time to prevent hacker attacks.

[0059] Step 7: Design an energy management module and a motor control module in the energy management layer. The energy management module monitors the battery, real-time monitors the battery status, and uses an active balancing method to extend the battery life. The motor control module real-time monitors the motor operation status and uses the vector control algorithm FOC to improve the motor efficiency.

[0060] Among them, the energy management module also dynamically adjusts the motor output power according to the road conditions and passenger capacity, uses a prediction algorithm to optimize the energy consumption strategy, intelligently schedules charging according to the battery status and trip plan, uses a linear programming algorithm to minimize the charging cost, and shortens the charging time. Fault warning includes health status monitoring, real-time monitors the health status of the battery and the motor (internal resistance, temperature), and uses a machine learning algorithm (SVM) to predict potential faults. Provide repair suggestions (such as replacing the battery, overhauling the motor), reduce vehicle downtime, and use fault tree analysis (FTA) to improve the fault tolerance of the system.

[0061] Step 8: Design an interaction module in the user interaction layer. The interaction module uses a face recognition model for face recognition to board the vehicle, and also provides QR code / NFC scanning to board the vehicle, and provides a multi-language interface, using the automatic speech recognition (ASR) method to support multi-language interaction.

[0062] The interaction module also uses a message queue to push bus information in real time, provides personalized services using a recommendation algorithm, and controls barrier-free facilities to serve special groups. The interaction module pushes information through touchscreens, cameras, and microphone devices.

[0063] Step 9: Design a security redundancy module in the security and redundancy layer. The security redundancy module monitors hardware failures, software anomalies, or external attacks, deploys redundant hardware, configures dual hardware, uses encryption technology for network data transmission, and uses distributed storage for data backup.

[0064] Among them, the security redundancy module uses encryption technology for network data transmission, uses the TLS / SSL protocol to encrypt the communication link to prevent data leakage, and uses digital certificates to authenticate both communication parties to prevent man-in-the-middle attacks;

[0065] Deploy a firewall and an intrusion detection system IDS to monitor network traffic in real time, and use blockchain technology to record communication logs to prevent data tampering.

[0066] It is also possible to gradually verify and optimize the unmanned intelligent bus system from simulation testing to large-scale deployment. First, conduct simulation testing to verify algorithms and system logic in a virtual environment, verify the feasibility and performance of the system architecture in a virtual environment, and discover and fix potential problems in the design. Secondly, conduct closed-road testing to gradually verify the coordination of the perception, decision-making, and execution modules, verify the actual performance of the system in a real but controlled environment, and ensure the safety and reliability of the system in a real scenario. Thirdly, conduct small-scale pilot testing, conduct trial operations on fixed routes, such as in industrial parks and airports, conduct small-scale trial operations in a real urban environment, and verify the adaptability and stability of the system in actual traffic. Finally, conduct large-scale deployment, deeply integrate with urban traffic management systems, such as intelligent traffic signals, and deploy the unmanned bus system on a large scale within the city to achieve the efficient operation and continuous optimization of the system.

[0067] Embodiment 2

[0068] The present invention also provides an unmanned intelligent bus system, including:

[0069] The sensor module, data fusion and analysis module, and fault tolerance module in the perception layer. The sensor module collects vehicle internal and external environment data in real time, provides information input for subsequent decision-making and execution. The data fusion and analysis module fuses the data collected by the sensor module, uses a deep learning convolutional neural network CNN to identify traffic signs and pedestrian intentions, uses a target detection algorithm to detect dynamic obstacles in real time. At the same time, the fault tolerance module cross-verifies the data of multiple sensor modules. When a certain sensor module fails, it instructs the data collected by other sensor modules to be used for data substitution.

[0070] The driving strategy planning module at the decision-making layer. The driving strategy planning module generates path planning and behavior decisions based on the data provided by the perception layer, and generates driving instructions according to the path planning and behavior decisions. The path planning includes global path planning and local path planning, and the behavior decision includes a rule engine. Preset traffic rules are used as decision boundaries, and a finite state machine (FSM) is used to manage the vehicle state.

[0071] The execution module at the execution layer. The execution module converts the driving instructions generated by the decision-making layer into actual actions of the vehicle.

[0072] The communication module at the communication layer. The communication module is responsible for data transmission between vehicle-to-vehicle (V2V), vehicle-to-infrastructure (V2I), and vehicle-to-cloud (V2C).

[0073] The cloud platform management module at the cloud platform layer. The cloud platform management module uses a distributed architecture for global scheduling, data analysis, and remote monitoring to achieve coordination among vehicles, passengers, and transportation infrastructure. It provides scheduling services and monitoring services using microservices, and performs local decision-making and data processing at edge computing nodes to reduce dependence on the cloud.

[0074] The energy management module and motor control module at the energy management layer. The energy management module monitors the battery and real-time battery status, and uses an active balancing method to extend the battery life. The motor control module real-time monitors the motor operating status and uses a vector control algorithm (FOC) to improve motor efficiency.

[0075] The interaction module at the user interaction layer. The interaction module uses a face recognition model for face recognition to board the vehicle, provides QR code / NFC scanning to board the vehicle, and provides a multi-language interface. It supports multi-language interaction using the automatic speech recognition (ASR) method.

[0076] The safety and redundancy module at the safety and redundancy layer. The safety and redundancy module monitors hardware failures, software anomalies, or external attacks, deploys redundant hardware, configures dual hardware, and uses encryption technology for network data transmission and distributed storage for data backup.

[0077] Regarding the information interaction and execution process among the above-mentioned modules in the system, since they are based on the same concept as the method embodiment of the present invention, the specific content can be referred to the description in the method embodiment of the present invention, and will not be elaborated here.

[0078] Similarly, the system of the present invention can improve the quality of public transportation services, reduce operating costs, and reduce carbon emissions. It can also promote the progress in the fields of autonomous driving, artificial intelligence, communication technology, and big data. At the same time, the driverless bus system will also promote policy innovation and the improvement of traffic management level, providing important support for the construction of smart cities. In the future, with the continuous progress of technology and the in-depth application, the driverless bus system will become an important part of urban transportation, providing a more convenient, efficient, and environmentally friendly travel mode for people.

[0079] It should be noted that not all steps and modules in the above-mentioned processes and system structures are necessary, and some steps or modules can be ignored according to actual needs. The execution order of each step is not fixed and can be adjusted according to needs. The system structures described in the above-mentioned embodiments can be physical structures or logical structures, that is, some modules may be implemented by the same physical entity, or some modules may be implemented separately by multiple physical entities, or some components in multiple independent devices may be jointly implemented.

[0080] The above-mentioned embodiments are only preferred embodiments given to fully illustrate the present invention, and the protection scope of the present invention is not limited thereto. Equivalent substitutions or transformations made by those skilled in the art on the basis of the present invention are all within the protection scope of the present invention. The protection scope of the present invention shall be subject to the claims.

Claims

1. A method for implementing an unmanned intelligent public transportation system, characterized in that include: Step 1: Design the architecture of the unmanned smart bus system in layers. The architecture layers include perception layer, decision layer, execution layer, communication layer, cloud platform layer, energy layer, user interaction layer, and security and redundancy layer. Step 2: Design sensor modules, data fusion analysis modules, and fault-tolerant modules at the perception layer. The sensor modules collect real-time data about the vehicle's internal and external environments to provide information input for subsequent decision-making and execution. The data fusion analysis module fuses the data collected by the sensor modules and uses a deep learning convolutional neural network (CNN) to identify traffic signs and pedestrian intentions. It uses a target detection algorithm to detect dynamic obstacles in real time. At the same time, the fault-tolerant module cross-validates the data from multiple sensor modules. When a sensor module fails, it instructs other sensor modules to collect data for data replacement. Step 3: Design a driving strategy planning module at the decision-making layer. The driving strategy planning module generates path planning and behavior decisions based on the data provided by the perception layer, and generates driving instructions based on path planning and behavior decisions. Path planning includes global path planning and local path planning. Behavior decisions include a rule engine, preset traffic rules as decision boundaries, and use finite state machines (FSMs) to manage vehicle status. Step 4: Design an execution module at the execution layer. The execution module converts the driving instructions generated by the decision layer into actual vehicle actions. Step 5: Design the communication module in the communication layer. The communication module is responsible for data transmission between vehicles (V2V), vehicles and infrastructure (V2I), and vehicles and the cloud (V2C). Step 6: Design a cloud platform management module at the cloud platform layer. The cloud platform management module uses a distributed architecture for global scheduling, data analysis, and remote monitoring to achieve coordination among vehicles, passengers, and transportation infrastructure. It uses microservices to provide scheduling and monitoring services, and performs local decision-making and data processing at edge computing nodes to reduce cloud dependency. Step 7: Design energy management module and motor control module at the energy management level. The energy management module monitors the battery, monitors the battery status in real time, and uses active balancing method to extend battery life. The motor control module monitors the motor operating status in real time and uses vector control algorithm FOC to improve motor efficiency. Step 8: Design an interaction module at the user interaction layer. The interaction module uses a face recognition model to perform face recognition boarding, provides QR code / NFC code scanning to board the bus, and provides a multi-language interface. It uses speech recognition ASR method to support multi-language interaction. Step 9: Design a safety redundancy module at the security and redundancy layer. The safety redundancy module monitors hardware failures, software anomalies, or external attacks, deploys redundancy for hardware, configures duplicate hardware, uses encryption technology for network data transmission, and uses distributed storage for data backup.

2. The method for implementing an unmanned intelligent public transportation system according to claim 1, characterized in that In step 2, the sensor module is configured with a laser radar, which is used to generate a 3D point cloud map to detect obstacles; the sensor module is configured with a multi-camera, which covers the area around the vehicle and identifies traffic signs, lane lines, pedestrians, and emergencies; the sensor module is configured with an ultrasonic sensor, which is used to avoid obstacles in close-range parking and low-speed scenarios; The data fusion analysis module uses Kalman filtering and particle filtering algorithms to fuse the data collected by lidar and multi-camera to generate a unified environmental model.

3. The method for implementing an unmanned intelligent public transportation system according to claim 1 or 2, characterized in that In step 3, the driving strategy planning module generates a path plan based on the data provided by the perception layer, including: using the A* algorithm and the Dijkstra algorithm to generate a global path plan. The global path plan is the optimal path from the starting point to the end point. The path is dynamically adjusted based on the map and real-time traffic conditions. The dynamic window method DWA or model predictive control MPC is used to generate a smooth local path. Behavioral decisions include collaborative decision-making, based on which the status of traffic lights and the location of other vehicles are obtained through C-V2X to control the driving of multiple vehicles in formation and reduce wind resistance and energy consumption.

4. The method for implementing an unmanned intelligent public transportation system according to claim 1, characterized in that In step 4, the execution module converts the driving instructions generated by the decision layer into actual actions of the vehicle, including: The execution module converts the vehicle's wire-controlled steering, adjusts the steering wheel angle according to the path planning instructions, configures the motor to drive the steering mechanism for steering, and uses the wire-controlled drive to adjust the vehicle speed according to the road conditions and traffic flow, and configures a high-performance motor controller for speed regulation. The execution module configures the motor control to use the vector control algorithm FOC to control the motor, dynamically adjust the motor output power according to road conditions; and recover kinetic energy during coasting or braking and store it in the battery.

5. The method for implementing an unmanned intelligent public transportation system according to claim 1, characterized in that In step 5, the communication module is responsible for data transmission between vehicles (V2V), vehicles and infrastructure (V2I), and vehicles and the cloud (V2C), including: The communication protocol and standard use V2X communication, the cellular vehicle-to-everything (C-V2X) protocol, and are compatible with the dedicated short-range communication (DSRC) protocol, and are interconnected with existing infrastructure. The communication module uses CAN bus and Ethernet high-speed communication protocols to achieve efficient data interaction among various subsystems in the vehicle, supports time-sensitive network TSN, and ensures real-time transmission of key data.

6. The method for implementing an unmanned intelligent public transportation system according to claim 1, characterized in that In step 6, the distributed file system HDFS is used at the cloud platform layer to store massive data, and the big data framework Hadoop or Spark is used to clean, analyze and mine the data, generate operation reports, and provide support for decision-making; Use global intelligent scheduling algorithms to optimize bus routes and schedules, and dynamically adjust scheduling strategies based on real-time traffic conditions and passenger flow data; Remote monitoring and real-time monitoring Use visualization tools such as Grafana or Kibana to monitor vehicle status and operation in real time and set up early warning mechanisms.

7. The method for implementing an unmanned intelligent public transportation system according to claim 1, characterized in that In step 7, the energy management module also dynamically adjusts the motor output power according to road conditions and passenger volume, uses a predictive algorithm to optimize the energy consumption strategy, intelligently schedules charging according to battery status and trip plan, and uses a linear programming algorithm to minimize charging costs and shorten charging time.

8. The method for implementing an unmanned intelligent public transportation system according to claim 1, characterized in that Step 8 The interactive module also uses the message queue to push bus information in real time, uses the recommendation algorithm to provide personalized services, and controls barrier-free facilities to serve special groups. The interactive module pushes information through touch screens, cameras, and microphone devices.

9. The method for implementing an unmanned intelligent public transportation system according to claim 1, The feature is that step 9: the security redundancy module uses encryption technology for network data transmission, uses TLS / SSL protocol to encrypt the communication link to prevent data leakage, and uses digital certificates to authenticate both parties in communication to prevent man-in-the-middle attacks; Deploy firewalls and intrusion detection systems (IDS) to monitor network traffic in real time, and use blockchain technology to record communication logs to prevent data tampering.

10. An unmanned intelligent public transportation system, characterized by include: The sensor module, data fusion analysis module, and fault-tolerant module of the perception layer. The sensor module collects real-time data on the vehicle's internal and external environment to provide information input for subsequent decision-making and execution. The data fusion analysis module fuses the data collected by the sensor module and uses the deep learning convolutional neural network CNN to identify traffic signs and pedestrian intentions. It uses the target detection algorithm to detect dynamic obstacles in real time. At the same time, the fault-tolerant module cross-validates the data of multiple sensor modules. When a sensor module fails, it instructs the data collected by other sensor modules to be used as data replacement. The driving strategy planning module of the decision-making layer generates path planning and behavior decisions based on the data provided by the perception layer, and generates driving instructions based on path planning and behavior decisions. Path planning includes global path planning and local path planning. Behavior decisions include rule engines, preset traffic rules as decision boundaries, and use finite state machines (FSMs) to manage vehicle status. The execution module of the execution layer converts the driving instructions generated by the decision layer into actual actions of the vehicle. The communication module of the communication layer is responsible for data transmission between vehicles (V2V), vehicles and infrastructure (V2I), and vehicles and the cloud (V2C). The cloud platform management module of the cloud platform layer uses a distributed architecture to perform global scheduling, data analysis, and remote monitoring to achieve the coordination of vehicles, passengers, and transportation infrastructure. It uses microservices to provide scheduling and monitoring services, and performs local decision-making and data processing at edge computing nodes to reduce cloud dependence. The energy management module and motor control module of the energy management layer monitor the battery, monitor the battery status in real time, and use active balancing methods to extend battery life. The motor control module monitors the motor operating status in real time and uses the vector control algorithm FOC to improve motor efficiency. The interactive module of the user interaction layer uses a face recognition model to board the bus, provides QR code / NFC code scanning, and provides a multi-language interface. It uses the voice recognition ASR method to support multi-language interaction. The safety redundancy module of the security and redundancy layer monitors hardware failures, software anomalies or external attacks, deploys redundancy for hardware, configures duplicate hardware, uses encryption technology for network data transmission, and uses distributed storage for data backup.

Images