An autonomous driving electronic and electrical system

Abstract

The application discloses an automatic driving electronic and electrical system, and relates to the technical field of automatic driving, which comprises a central dispatching manager, a perception module, a decision module, a control module, a plurality of first computing nodes and a plurality of edge computing nodes; the perception module, the decision module, the control module and the plurality of first computing nodes are connected with the central dispatching manager; each edge computing node is connected with a vehicle-mounted sensor, the edge computing node is used for preprocessing vehicle-mounted sensor data and sending the preprocessed data to the perception module; each first computing node is used for executing a specific instruction, the specific instruction comprises a resource allocation instruction generated by the central dispatching manager according to a real-time computing resource state, and the resource allocation instruction is used for allocating computing resources to the perception module, the decision module and the control module. The application reduces the maintenance and expansion difficulty of the system.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of autonomous driving, and in particular to an autonomous driving electronic and electrical system. BACKGROUND

[0002] With the rapid development of autonomous driving technology, the demand for real-time data processing capability of vehicles has significantly increased. The existing centralized computing architecture is difficult to efficiently process a large amount of sensor data in the autonomous driving scenario, and the system reliability and scalability are greatly limited. In addition, with the continuous iteration of new technologies, the function modules in the autonomous driving system need to be frequently upgraded and replaced, and the difficulty of maintenance and expansion of the existing system needs to be reduced. SUMMARY

[0003] The purpose of the present application is to provide an autonomous driving electronic and electrical system, which reduces the difficulty of maintenance and expansion of the system.

[0004] To achieve the above purpose, the present application provides the following solutions:

[0005] The present application provides an autonomous driving electronic and electrical system, comprising: a central scheduling manager, a perception module, a decision module, a control module, a plurality of first computing nodes and a plurality of edge computing nodes; each of the edge computing nodes is connected with the perception module, the perception module is connected with the decision module, and the decision module is connected with the control module; the perception module, the decision module, the control module and a plurality of the first computing nodes are connected with the central scheduling manager;

[0006] Each of the edge computing nodes is connected with a vehicle-mounted sensor, and the edge computing node is used for preprocessing the vehicle-mounted sensor data and sending the preprocessed data to the perception module;

[0007] The perception module is used for fusing the preprocessed data corresponding to a plurality of vehicle-mounted sensors, generating a three-dimensional environment model, and sending the three-dimensional environment model to the decision module and the central scheduling manager;

[0008] The decision module is used for sequentially performing path planning, behavior prediction and generating a driving strategy according to the three-dimensional environment model, and sending the driving strategy to the control module;

[0009] The control module is used for generating a vehicle control command according to the driving strategy;

[0010] Each of the first computing nodes is used for executing a specific instruction, and the specific instruction comprises a resource allocation instruction generated by the central scheduling manager according to the real-time computing resource state, and the resource allocation instruction is used for allocating computing resources to the perception module, the decision module and the control module.

[0011] Optionally, the autonomous driving electronic and electrical system further comprises a plurality of second computing nodes, each of the second computing nodes is connected with one of the functional modules, the functional modules include the perception module, the decision module and the control module.

[0012] Each of the second computing nodes is configured to perform a computing task corresponding to one of the functional modules.

[0013] Optionally, the autonomous driving electronic and electrical system further comprises a plurality of third computing nodes, the third computing nodes are configured to perform auxiliary computing tasks, the auxiliary computing tasks include data storage, map updating, remote communication and system log recording.

[0014] Optionally, the central scheduling manager is further configured to monitor a load condition of each of the computing nodes, and perform task scheduling on each of the computing nodes according to the load condition of each of the computing nodes, the computing nodes include the first computing nodes, the second computing nodes and the third computing nodes; each of the computing nodes communicates with the central scheduling manager through a high-speed communication network.

[0015] Optionally, the vehicle-mounted sensors include a camera, a laser radar, a millimeter wave radar and an ultrasonic radar.

[0016] The sensor data collected by the camera is image data, the sensor data collected by the laser radar is three-dimensional point cloud data, the sensor data collected by the millimeter wave radar is a relative distance and a speed of a detected object and a vehicle, and the sensor data collected by the ultrasonic radar is a distance of an object within a set range.

[0017] Optionally, the edge computing node comprises a data preprocessing unit, the data preprocessing unit is configured to filter, down-sample and edge detect the image data, and de-noise and object identify the three-dimensional point cloud data.

[0018] Optionally, the perception module comprises a three-dimensional environment model generation unit, the three-dimensional environment model generation unit is configured to perform feature extraction and feature fusion on the preprocessed image data and the preprocessed three-dimensional point cloud data by using a deep learning method, and generate a three-dimensional environment model.

[0019] Optionally, the central scheduling manager is further configured to determine whether to generate an obstacle avoidance driving strategy according to the relative distance and the speed of the detected object and the vehicle, and the distance of the object within the set range.

[0020] Optionally, the decision module and the control module communicate through a high-speed communication network.

[0021] Optionally, the vehicle control commands include steering, acceleration, deceleration, and braking.

[0022] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application provides an autonomous driving electronic and electrical system, which realizes the modularization of the system architecture through a central dispatch manager, a perception module, a decision module, a control module, multiple first computing nodes and multiple edge computing nodes, enabling rapid replacement and upgrading of each functional module, and reducing the difficulty of system maintenance and expansion. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of the structure of an automatic driving electronic and electrical system provided in one embodiment of this application. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0026] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0027] This application provides an autonomous driving electronic and electrical system, such as Figure 1 As shown, the autonomous driving electronic and electrical system includes: a central dispatch manager, a perception module, a decision module, a control module, multiple first computing nodes, and multiple edge computing nodes; each edge computing node is connected to the perception module, the perception module is connected to the decision module, and the decision module is connected to the control module; the perception module, the decision module, the control module, and the multiple first computing nodes are all connected to the central dispatch manager.

[0028] The autonomous driving electronic and electrical system further includes: multiple second computing nodes, each second computing node being connected to a functional module, the functional module including the perception module, the decision module and the control module.

[0029] Each of the second computing nodes is configured to perform a computing task corresponding to each of the functional modules.

[0030] The autonomous driving electronic and electrical system further comprises a plurality of third computing nodes configured to perform auxiliary computing tasks, including data storage, map updating, remote communication, and system logging.

[0031] The central scheduling manager is the core control unit of the system, responsible for task allocation, resource management, and coordination of the entire system. The central scheduling manager monitors the running state of each module and the load of each computing node, adjusts the allocation of tasks in real time, and ensures the efficient operation of each functional module. The central scheduling manager is connected to each functional module (such as the perception module, decision module, and control module) through a high-speed communication network. It is also connected to all distributed computing nodes for allocating computing tasks, delivering instructions, and monitoring system operation.

[0032] The central scheduling manager is further configured to monitor the load of each of the computing nodes, and to perform task scheduling for each of the computing nodes according to the load of each of the computing nodes, the computing nodes including first computing nodes, second computing nodes, and third computing nodes; each of the computing nodes communicates with the central scheduling manager through a high-speed communication network.

[0033] For example, when the system enters complex road conditions, the central scheduling manager will preferentially allocate more computing resources to the decision module to ensure that path planning and obstacle avoidance algorithms can be completed quickly. If a computing node is overloaded or fails, the central scheduling manager can quickly migrate tasks to other redundant nodes to ensure high reliability and stability of the system.

[0034] The perception module is configured to fuse pre-processed data from multiple vehicle-mounted sensors to generate a three-dimensional environmental model, and to send the three-dimensional environmental model to the decision module and the central scheduling manager.

[0035] The vehicle-mounted sensors include cameras, lidar, millimeter wave radar, and ultrasonic radar. Vehicle-mounted sensors capture information in the external environment in real time. The signals generated by these sensors contain data about the surrounding environment, such as roads, pedestrians, and other vehicles, and also collect the vehicle's own state signals, such as speed, steering angle, acceleration, and braking status. This information serves as the basic data input for the system's operation.

[0036] Each of the edge computing nodes is connected to a vehicle-mounted sensor and located near the vehicle-mounted sensor. The edge computing nodes are configured to pre-process and perform simple target recognition on the vehicle-mounted sensor data, and to send the pre-processed data to the perception module, thereby reducing computing pressure and improving the real-time performance of the system.

[0037] Each of the first computing nodes is configured to execute a specific instruction, and the specific instruction includes a resource allocation instruction generated by the central scheduling manager according to a real-time computing resource state, and the resource allocation instruction is used for allocating computing resources to the perception module, the decision module and the control module. The first computing node dynamically allocates task loads according to the instruction of the central scheduling manager, and undertakes the computing work of the perception module, the decision module and the control module.

[0038] The functional modules are connected or communicate through standardized interfaces.

[0039] The application realizes quick replacement and upgrade of the functional modules through the standardized interfaces, and optimizes system performance by using flexible scheduling of distributed computing resources. The architecture not only improves the reliability and expansibility of the system, but also enhances the real-time performance and flexibility of the autonomous driving system in dealing with complex road conditions.

[0040] The sensor data collected by the camera is image data, the sensor data collected by the laser radar is three-dimensional point cloud data, the sensor data collected by the millimeter wave radar is the relative distance and speed of the detected object and the vehicle, and the sensor data collected by the ultrasonic radar is the distance of the object within the set range.

[0041] The edge computing node includes a data preprocessing unit; the data preprocessing unit is configured to filter, downsample and edge detect the image data; and denoise and object identify the three-dimensional point cloud data.

[0042] The preprocessing of the sensor data collected by the millimeter wave radar by the data preprocessing unit includes noise filtering, target detection and preliminary classification, distance and speed estimation, and target tracking.

[0043] 1. Noise filtering: millimeter wave radars are susceptible to noise, such as interference from other radar systems, environmental reflections, etc. The edge computing node will first perform a noise filtering algorithm to remove invalid or unreliable signals, ensuring the accuracy of subsequent processing.

[0044] 2. Target detection and preliminary classification: millimeter wave radars are used to detect the relative distance and speed of objects. The edge computing node processes the radar echo signals, detects targets, and performs preliminary classification, such as identifying moving objects (such as vehicles, pedestrians) or stationary objects (such as roadblocks). Preliminary classification helps reduce the computational burden of the central system.

[0045] 3. Distance and speed estimation: millimeter wave radars measure the relative speed of objects through the Doppler effect. The edge node is responsible for quickly processing the Doppler shift signals to estimate the speed and distance of the target object, and provides accurate input data for the subsequent decision module.

[0046] 4. Object Tracking: The edge computing nodes will perform simple object tracking algorithms to track multiple detected objects, determine their relative motion trajectories, and facilitate subsequent dynamic planning and obstacle avoidance.

[0047] The data preprocessing unit preprocesses the sensor data collected by the ultrasonic sensor, including signal denoising and filtering, distance estimation and object detection, and obstacle size and type inference.

[0048] 1. Signal Denoising and Filtering: The signal of the ultrasonic sensor is easily disturbed by environmental factors such as air flow and humidity. The edge computing node will first perform signal denoising to filter out false echoes or environmental disturbances and improve data accuracy.

[0049] 2. Distance Estimation and Object Detection: The ultrasonic sensor is mainly used for detecting objects at close range. The edge node will process the echo time to accurately estimate the distance of the object and output the detection result in real time, ensuring that the system can accurately perceive the position of obstacles in low-speed environments or during parking.

[0050] 3. Obstacle Size and Type Inference: The edge computing node can infer the size and possible type of obstacles (such as vehicles, walls, and other objects) by analyzing the reflection intensity and echo characteristics of ultrasonic signals, providing further information support for the decision-making module.

[0051] The preprocessing of the millimeter wave radar mainly focuses on noise filtering, target detection, distance and speed estimation, and target tracking to ensure that the system can understand the dynamic object situation around the vehicle in real time.

[0052] The preprocessing of the ultrasonic sensor focuses on signal denoising, close-range obstacle detection, and obstacle feature inference, which is usually used for precise perception in parking scenarios and low-speed driving.

[0053] The perception module includes a data acquisition unit and a three-dimensional environment model generation unit. The data acquisition unit is used to receive data transmitted by the edge computing node, and the three-dimensional environment model generation unit uses deep learning methods to extract and fuse features from preprocessed image data and preprocessed three-dimensional point cloud data to generate a three-dimensional environment model. The data processing in this stage aims to provide the system with an overall and accurate model of the surrounding environment.

[0054] The decision module is used to sequentially perform path planning, behavior prediction and generate driving strategy according to the three-dimensional environment model, and send the driving strategy to the control module. The driving strategy includes the steering, acceleration, deceleration and obstacle avoidance strategy of the vehicle. Specifically, first, based on the current road conditions and vehicle position, a path planning algorithm is run to generate the optimal driving path for the vehicle. At the same time, through a behavior prediction algorithm, the motion trajectory of other dynamic objects (such as pedestrians, vehicles) is predicted, and the necessary obstacle avoidance strategy is executed. In this process, the system will comprehensively consider the current vehicle speed, steering angle and obstacle distribution around the vehicle to ensure that the generated driving strategy is safe and efficient.

[0055] The control module receives the control commands generated by the decision module and is specifically responsible for the action execution of the vehicle, including steering, acceleration, braking and other operations. It is the execution layer of the vehicle interacting with the external environment, directly controlling the power and control system of the vehicle. Specifically: after the decision module completes the path planning, it will generate specific control instructions, such as acceleration, deceleration, steering and braking. Subsequently, these instructions are transmitted to the control module through a high-speed communication network. The control module is responsible for converting these instructions into actual vehicle operations. The control module is connected with the steering system, brake system and power system of the vehicle, and directly controls these hardware devices through the electronic control unit (ECU). For example, when the decision module requires the vehicle to turn right, the control module will transmit the instruction to the steering system to adjust the steering wheel angle. Similarly, if a front obstacle is detected and emergency braking is required, the control module will immediately activate the braking system to ensure the vehicle stops safely.

[0056] The autonomous driving electronic and electrical system further comprises a high-speed communication network, such as an Ethernet, which is used to ensure real-time communication between modules and nodes, especially high-precision, low-latency data transmission required during autonomous driving, with time synchronization, flow control and delay guarantee functions to ensure that data interaction between different modules within the system can be completed within the predetermined time window. The central dispatching manager, the perception module, the decision module, the control module and the edge computing node are tightly connected through the high-speed communication network to ensure that data and instructions can be transmitted within milliseconds.

[0057] The first computing node, the second computing node and the third computing node are three kinds of computing nodes. The third computing node is an independent processing unit distributed in different positions and is responsible for processing specific tasks.

[0058] The first computing node dynamically allocates task load according to the instructions of the central dispatching manager and undertakes the computing work of the perception, decision and control modules. More specifically, the role of the first computing node specifically includes: resource allocation and load balancing, task scheduling execution and high-priority task processing.

[0059] The first computing node accepts resource allocation instructions from the central scheduling manager, performs system-level task allocation and load balancing. The first computing node is responsible for dynamically allocating resources and performing tasks according to real-time computing needs.

[0060] The central scheduling manager issues tasks through the first computing node, and the first computing node performs global task scheduling. For example, emergency obstacle avoidance tasks and complex path planning tasks can be efficiently processed on these nodes.

[0061] The first computing node is usually used to process high-priority tasks, which have a direct impact on system safety, such as the computing and processing of perception modules and decision modules.

[0062] The first computing node is usually located in the central control unit of the vehicle (such as the domain controller or central processing unit), and is concentrated on the main computing platform of the vehicle to achieve fast communication and task allocation.

[0063] The second computing node is directly connected to specific functional modules (such as perception modules, decision modules, control modules, etc.), and is responsible for processing local tasks of the module. They perform specific function processing according to the instructions of the central scheduling manager or the needs of the task module. The role of the second computing node includes local computing and data processing, data preprocessing and fusion, task execution and feedback.

[0064] Local computing and data processing: The second computing node is responsible for processing data related to the functional module it is connected to. For example, the second computing node connected to the perception module is responsible for preliminary processing of data from lidar, camera, etc. sensors; the second computing node connected to the decision module is responsible for computing path planning, obstacle avoidance, etc.

[0065] Data preprocessing and fusion: The second computing node of the perception module will perform sensor data preprocessing and data fusion, such as comprehensive processing of camera, lidar, millimeter wave radar data, and generating a preliminary environment model.

[0066] Task execution and feedback: The second computing node is responsible for executing computing tasks related to its functional module and feeding back the results to the central scheduling manager. For example, the second computing node of the control module is responsible for executing specific control instructions (such as steering, acceleration, braking), and reporting task status to the scheduling manager.

[0067] The second computing node is set in the distributed controller location: It can be located in multiple locations of the vehicle, and is responsible for processing tasks related to perception, decision, and control, reducing the performance bottleneck caused by centralized computing.

[0068] The third computing nodes are not directly connected to the central scheduling manager or specific functional modules, but independently handle certain tasks or provide auxiliary computing power for the system. They are usually used to handle tasks that do not require real-time performance or do not directly affect the core decisions of the system. The third computing nodes are used to undertake some non-core, low-priority tasks that do not require real-time response but are crucial for the long-term stability and data accumulation of the system. The third computing nodes are also used as redundant backups for the system, taking over part of the tasks when the main computing nodes fail, to ensure the fault tolerance of the system. The third computing nodes are also used to handle remote communication tasks, responsible for communication with external cloud platforms or other vehicles, for data exchange and sharing.

[0069] The third computing nodes are usually deployed in positions that do not affect the main sensor layout, such as the roof, trunk, or other non-core control areas. These positions facilitate the expansion of computing resources and facilitate the connection with external networks.

[0070] The control module is configured to generate vehicle control commands according to the driving strategy.

[0071] The central scheduling manager is further configured to determine whether to generate an obstacle avoidance driving strategy according to the relative distance and speed of the detected object and the vehicle, and the distance of the object within the set range.

[0072] The decision module and the control module communicate through a high-speed communication network.

[0073] The vehicle control commands include steering, acceleration, deceleration, and braking.

[0074] The dynamic allocation of computing resources and task scheduling for the central scheduling manager specifically includes:

[0075] 1. Real-time monitoring of system status: The central scheduling manager continuously monitors the real-time status of each computing node (specifically, central processing units, dedicated controllers, etc.) and each edge computing node. The monitoring indicators include CPU and GPU usage, memory occupancy, network delay, task queue length, etc. At the same time, the system also monitors each functional module (such as the perception module, decision module, control module) to view the task amount, task priority, and task complexity of each functional module. Through these monitoring, the central scheduling manager can dynamically obtain the current resource usage and task load, which serves as the basis for subsequent resource scheduling.

[0076] 2. Task priority analysis: Tasks in the autonomous driving system are prioritized based on their importance and real-time requirements. For example, emergency obstacle avoidance tasks have the highest priority, while path planning, map updating, and other tasks have lower priorities. The central scheduling manager allocates resources based on the priority and latency requirements of tasks. For example, high-priority tasks (such as obstacle avoidance calculations) are immediately assigned to the most idle or best-suited computing nodes, while low-priority tasks (such as data storage or background map updates) are delayed or assigned to lower-performance nodes.

[0077] 3. Dynamic allocation of computing resources: The central scheduling manager dynamically selects appropriate computing nodes to execute tasks based on the real-time status of each computing node (such as available resources, communication bandwidth, processing power, etc.) and the computational complexity of tasks. Different computing nodes may have strengths, such as some nodes being good at parallel processing (suitable for processing lidar data) and others being better suited for image processing (suitable for processing camera data). To avoid overloading computing nodes, the central scheduling manager implements load balancing strategies. When the load of a computing node reaches a certain threshold, the scheduling manager automatically assigns new tasks to lower-load nodes, preventing performance bottlenecks in individual nodes from affecting the response time of the entire system. When task priorities change or the volume of tasks suddenly increases, the central scheduling manager can dynamically adjust resource allocation. For example, when the system detects a complex scenario or an emergency situation (such as an obstacle ahead that needs to be avoided), the scheduling manager immediately allocates more computing resources to the decision-making module, ensuring that the obstacle avoidance strategy can be quickly generated.

[0078] 4. Task scheduling and migration: The central scheduling manager typically employs various scheduling strategies, such as earliest deadline first (EDF), least slack time first, or round-robin scheduling. These strategies help the system decide which task should be executed first, ensuring that time-sensitive tasks are completed on time. When a computing node becomes overloaded, fails, or needs to be maintained, the central scheduling manager can migrate the tasks being executed to other available nodes. This process is achieved through virtualization or containerization technologies, allowing smooth migration without interrupting task execution. For example, tasks can be migrated from central nodes to edge computing nodes or between different edge nodes. The central scheduling manager also reserves a portion of computing resources for handling unexpected tasks. For example, when a vehicle is expected to enter complex road conditions, the central scheduling manager allocates more computing resources to the perception and decision-making modules in advance to handle the increased computational demand.

[0079] The application is an automatic driving electronic and electrical system, specifically a schematic diagram of an automatic driving electronic and electrical system based on a modular distributed computing architecture. The specific workflow is as follows: when the vehicle starts, the central dispatch manager first monitors the system state and collects real-time data from each module; the perception module preprocesses data from multiple sensors through the edge node, and then transmits the fused data to the decision module; the decision module generates driving strategies and control commands based on these data, and then executes specific vehicle operations through the control module. Throughout the process, the central dispatch manager allocates and adjusts the task load of each computing node in real time to ensure efficient and reliable operation of the system under different working conditions.

[0080] The modular design of the application allows flexible replacement and upgrade of the decision unit according to different algorithms or driving strategies. The control module is responsible for outputting control instructions such as steering, acceleration, and braking of the vehicle, and operates independently of other modules to ensure real-time control response in a distributed system. Distributed computing and task scheduling can reduce communication delay and improve real-time performance. The central dispatch manager allocates computing tasks in real time based on the load of the computing nodes and the current road conditions. The tasks of the perception module can be preferentially allocated to the edge nodes close to the sensors, while the tasks of the decision module and the control module are handled by the central nodes with stronger computing power. The scheduling algorithm can dynamically adjust the task allocation of the nodes according to the real-time performance requirements of the system to ensure optimal performance.

[0081] Each module of the application has a redundant design, that is, when a module or node fails, the system can quickly switch to a backup node to ensure uninterrupted function. In addition, through the self-diagnosis system, each module can be monitored in real time, and when an anomaly is detected, the system can self-repair or switch to a prepared computing node to ensure high availability of the automatic driving system.

[0082] Each module of the application uses a standardized interface, and each module can be quickly replaced or upgraded according to the latest technology without the need to reconstruct the entire system. For example, the perception module can improve perception accuracy by replacing new sensor hardware or upgrading algorithms. This design simplifies the maintenance process of the system and reduces the cost and complexity of later upgrades.

[0083] The application efficiently processes various signals collected by vehicle-mounted sensors through a distributed architecture, and utilizes modular design to enable independent operation and dynamic collaboration of each functional module. The collection, processing, decision-making and execution of signals are closely combined, and the central dispatch manager ensures reasonable task distribution and optimized resource use. Finally, the safe and reliable operation of the autonomous vehicle in complex environments is realized, with high real-time performance, high reliability and flexibility. Through this modular distributed computing architecture, the system not only meets the high computing demand of autonomous driving, but also improves the reliability, scalability and flexibility. The application can effectively cope with the rapid changes of future autonomous driving technology and provide efficient vehicle control and environmental perception capabilities.

[0084] The technical features of the above embodiments can be combined in any manner. To make the description concise, not all possible combinations of the technical features in the above embodiments are described, but as long as the combinations of the technical features do not exist contradictions, they should be considered as the scope of the present disclosure.

[0085] The principles and implementation modes of the application are described by applying specific examples herein, and the above embodiment descriptions are only used to help understand the method of the application and its core idea; meanwhile, for those skilled in the art, according to the idea of the application, the specific implementation modes and application ranges will have changes. In conclusion, the content of the specification should not be understood as a limitation of the application.

Claims

1. An autonomous driving electronic and electrical system, characterized in that, The automatic driving electronic and electrical system comprises a central scheduling manager, a perception module, a decision module, a control module, a plurality of first computing nodes and a plurality of edge computing nodes; each of the edge computing nodes is connected with the perception module, the perception module is connected with the decision module, and the decision module is connected with the control module; the perception module, the decision module, the control module and the plurality of first computing nodes are connected with the central scheduling manager; Each of the edge computing nodes is connected with a vehicle-mounted sensor, and the edge computing nodes are used for preprocessing vehicle-mounted sensor data and sending the preprocessed data to the perception module; The perception module is used for fusing the preprocessed data corresponding to a plurality of vehicle-mounted sensors, generating a three-dimensional environment model, and sending the three-dimensional environment model to the decision module and the central scheduling manager; The decision module is used for sequentially performing path planning, behavior prediction and generating a driving strategy according to the three-dimensional environment model, and sending the driving strategy to the control module; The control module is used for generating a vehicle control command according to the driving strategy; Each of the first computing nodes is used for executing a specific instruction, and the specific instruction comprises a resource allocation instruction generated by the central scheduling manager according to a real-time computing resource state, and the resource allocation instruction is used for allocating computing resources to the perception module, the decision module and the control module; The automatic driving electronic and electrical system further comprises a plurality of second computing nodes, each of which is connected with a functional module, and the functional module comprises the perception module, the decision module and the control module; Each of the second computing nodes is used for executing a computing task corresponding to each of the functional modules; The automatic driving electronic and electrical system further comprises a plurality of third computing nodes, and the third computing nodes are used for executing auxiliary computing tasks, and the auxiliary computing tasks comprise data storage, map updating, remote communication and system log recording.

2. The autonomous driving electronic electrical system of claim 1, wherein, The central scheduling manager is further used for monitoring the load conditions of each of the computing nodes, and performing task scheduling on each of the computing nodes according to the load conditions of each of the computing nodes, and the computing nodes comprise the first computing nodes, the second computing nodes and the third computing nodes; each of the computing nodes communicates with the central scheduling manager through a high-speed communication network.

3. The autonomous driving electronic electrical system of claim 1, wherein, The vehicle-mounted sensors comprise a camera, a laser radar, a millimeter wave radar and an ultrasonic radar; The sensor data collected by the camera is image data, the sensor data collected by the laser radar is three-dimensional point cloud data, the sensor data collected by the millimeter wave radar is the relative distance and speed of the detected object and the vehicle, and the sensor data collected by the ultrasonic radar is the distance of the object within a set range.

4. The autonomous driving electronic electrical system of claim 3, wherein, The edge computing node comprises a data preprocessing unit; the data preprocessing unit is used for filtering, downsampling and edge detection on the image data; and noise reduction and object recognition are performed on the three-dimensional point cloud data.

5. The autonomous driving electronic and electrical system according to claim 4, characterized in that The perception module comprises a three-dimensional environment model generation unit, which extracts and fuses features of the preprocessed image data and the preprocessed three-dimensional point cloud data by using a deep learning method to generate a three-dimensional environment model.

6. The autonomous driving electronic and electrical system according to claim 4, characterized in that, The central scheduling manager is further configured to determine whether to generate an obstacle avoidance driving strategy according to a relative distance and speed of the detected object and the vehicle, and a distance of the object within a set range.

7. The autonomous driving electronic electrical system of claim 1, wherein, The decision module and the control module communicate through a high-speed communication network.

8. The autonomous driving electronic electrical system of claim 1, wherein, The vehicle control command comprises steering, acceleration, deceleration and braking.

Images