Unmanned vehicle intelligent control method and system based on satellite radar

Abstract

The invention discloses an intelligent control method and system for an unmanned vehicle based on a satellite radar, and relates to the technical field of satellite radars, and the method comprises the steps: building a management and control system of the satellite radar based on each induction event, the working data of each radar module, and an interaction event between the radar modules; the dynamic control content of the unmanned vehicle is determined based on the management and control system of the satellite radar and the unmanned driving mode of the unmanned vehicle, the intelligent control event of the unmanned vehicle is determined according to the dynamic control content of the unmanned vehicle, the surrounding environment detected by the unmanned vehicle and the working process of the satellite radar, and the accuracy of the intelligent control event of the unmanned vehicle is improved. Therefore, the current control instruction of the unmanned vehicle is determined according to the intelligent control state of the unmanned vehicle, the driving scene where the unmanned vehicle is located and the previous control event of the unmanned vehicle, and the corresponding intelligent driving content is determined based on the current control instruction of the unmanned vehicle, the environment data detected by the satellite radar and the obstacle data. And the accuracy of the intelligent driving content is improved.

Description

Technical Field

[0001] This invention relates to the technical field of satellite radar, and in particular to an intelligent control method and system for unmanned vehicles based on satellite radar. Background Technology

[0002] With the development of technology, driverless cars are gradually being applied to people's lives and are driving autonomously on roads. Driverless cars are equipped with multiple radars, each of which detects external objects. In the current technology, a corresponding satellite radar is built based on the interaction of various radars to control the surrounding environment of the driverless car. The driverless car drives according to preset instructions, ignoring the current control instructions of the driverless car and the obstacle data detected by the satellite radar, which affects the accuracy of intelligent driving. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides an intelligent control method and system for unmanned vehicles based on satellite radar.

[0004] This invention provides an intelligent control method for unmanned vehicles based on satellite radar, comprising: Multiple radar modules are marked based on the detection of the unmanned vehicle, and the radar module sensing system is determined according to the distribution location of the multiple radar modules, their corresponding detection range, and the current position of the unmanned vehicle. Multiple sensing events are identified based on the radar module sensing system, and a satellite radar control system is constructed based on each sensing event, the working data of each radar module, and the interaction events between each radar module. The dynamic control content of the unmanned vehicle is determined based on the satellite radar control system and the unmanned driving mode of the unmanned vehicle. The intelligent control events of the unmanned vehicle are determined based on the dynamic control content of the unmanned vehicle, the surrounding environment detected by the unmanned vehicle, and the working history of the satellite radar. In the intelligent control events of autonomous vehicles, multiple intelligent control items are identified based on the recognition of the intelligent control events of autonomous vehicles. The intelligent control status of autonomous vehicles is determined according to the content of each intelligent control item, the corresponding priority of the item, and the driving data of autonomous vehicles. The system collects data on the driving scenario in which the autonomous vehicle is located. Based on the autonomous vehicle's intelligent control status, the driving scenario, and past control events, it determines the current control command for the autonomous vehicle. Based on the current control command, environmental data detected by satellite radar, and obstacle data, it determines the corresponding intelligent driving content.

[0005] This invention provides an intelligent control system for an unmanned vehicle based on satellite radar. This intelligent control system is applied to the aforementioned intelligent control method for an unmanned vehicle based on satellite radar. The intelligent control system for the unmanned vehicle based on satellite radar includes: The radar module sensing system module is used to mark multiple radar modules based on the detection of unmanned vehicles. The radar module sensing system is determined according to the distribution of multiple radar modules, their corresponding detection ranges, and the current position of the unmanned vehicle. The control system module is used to identify multiple sensing events based on the radar module's sensing system, and to construct a control system for the satellite radar based on each sensing event, the working data of each radar module, and the interaction events between each radar module. The intelligent control event module is used to determine the dynamic control content of the unmanned vehicle based on the satellite radar control system and the unmanned driving mode of the unmanned vehicle. The module determines the intelligent control events of the unmanned vehicle based on the dynamic control content of the unmanned vehicle, the surrounding environment detected by the unmanned vehicle, and the working history of the satellite radar. The intelligent control status module is used to identify multiple intelligent control items based on the intelligent control events of the autonomous vehicle, and to determine the intelligent control status of the autonomous vehicle based on the content of each intelligent control item, the corresponding priority of the item, and the driving data of the autonomous vehicle. The intelligent driving content module is used to collect the driving scene where the unmanned vehicle is located, determine the current control command of the unmanned vehicle based on the intelligent control status of the unmanned vehicle, the driving scene where the unmanned vehicle is located, and the previous control events of the unmanned vehicle, and determine the corresponding intelligent driving content based on the current control command of the unmanned vehicle, environmental data detected by satellite radar and obstacle data.

[0006] Compared with the prior art, the beneficial effects of the present invention are: In this embodiment of the invention, the dynamic control content of the unmanned vehicle is determined based on the satellite radar control system and the unmanned driving mode of the unmanned vehicle. The intelligent control events of the unmanned vehicle are determined based on the dynamic control content of the unmanned vehicle, the surrounding environment detected by the unmanned vehicle, and the working history of the satellite radar. A radar module sensing system is introduced to control the satellite radar control system. This approach takes into account the overall dynamic control content of the unmanned vehicle, the surrounding environment detected by the unmanned vehicle, and the working history of the satellite radar, thereby improving the accuracy of the intelligent control events of the unmanned vehicle.

[0007] Therefore, in the intelligent control events of autonomous vehicles, multiple intelligent control items are determined based on the identification of these events. The intelligent control state of the autonomous vehicle is determined according to the content, priority, and driving data of each item. The driving scene of the autonomous vehicle is collected, and the current control command of the autonomous vehicle is determined based on its intelligent control state, the driving scene, and its past control events. Based on the current control command, environmental data detected by satellite radar, and obstacle data, the corresponding intelligent driving content is determined. The intelligent control state of the autonomous vehicle is introduced, further controlling the current control command. This allows for a holistic consideration of the current control command, environmental data detected by satellite radar, and obstacle data, thus improving the accuracy of the intelligent driving content. Attached Figure Description

[0008] Figure 1 This is a flowchart illustrating the intelligent control method for an unmanned vehicle based on satellite radar in an embodiment of the present invention. Figure 2 This is a flowchart illustrating step S11 in the intelligent control method for unmanned vehicles based on satellite radar in this embodiment of the invention. Figure 3 This is a flowchart illustrating step S12 in the intelligent control method for unmanned vehicles based on satellite radar in this embodiment of the invention. Figure 4 This is a flowchart illustrating step S13 in the intelligent control method for unmanned vehicles based on satellite radar in this embodiment of the invention. Figure 5 This is a flowchart illustrating step S14 in the intelligent control method for unmanned vehicles based on satellite radar in this embodiment of the invention. Figure 6 This is a flowchart illustrating step S15 in the intelligent control method for unmanned vehicles based on satellite radar in this embodiment of the invention. Figure 7 This is a schematic diagram of the structural composition of the intelligent control system for an unmanned vehicle based on satellite radar in an embodiment of the present invention. Detailed Implementation

[0009] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0010] Please see Figures 1 to 7 An intelligent control method for unmanned vehicles based on satellite radar is proposed and applied to satellite radar scenarios. The intelligent control method for unmanned vehicles based on satellite radar includes: Step S11: Mark multiple radar modules based on the detection of the unmanned vehicle, and determine the radar module sensing system according to the distribution location of the multiple radar modules, their corresponding detection range and the current position of the unmanned vehicle; Step S12: Based on the identification of the radar module sensing system, determine multiple sensing events, and construct a satellite radar control system based on each sensing event, the working data of each radar module, and the interaction events between each radar module; Step S13: Determine the dynamic control content of the unmanned vehicle based on the satellite radar control system and the unmanned driving mode of the unmanned vehicle, and determine the intelligent control events of the unmanned vehicle based on the dynamic control content of the unmanned vehicle, the surrounding environment detected by the unmanned vehicle, and the working history of the satellite radar. Step S14: In the intelligent control events of the unmanned vehicle, multiple intelligent control items are determined based on the identification of the intelligent control events of the unmanned vehicle, and the intelligent control status of the unmanned vehicle is determined according to the item content, corresponding item priority and driving data of the unmanned vehicle. Step S15: Collect the driving scene where the unmanned vehicle is located, determine the current control command of the unmanned vehicle based on the intelligent control status of the unmanned vehicle, the driving scene where the unmanned vehicle is located, and the previous control events of the unmanned vehicle, and determine the corresponding intelligent driving content based on the current control command of the unmanned vehicle, the environmental data and obstacle data detected by satellite radar.

[0011] refer to Figure 2 In step S11, the specific steps are as follows: S111: Real-time monitoring of unmanned vehicles and detection of unmanned vehicles; determination of multiple radar signals for different paths based on the detection of unmanned vehicles; determination of corresponding radar modules based on the tracing of each radar signal; marking of multiple radar modules. S112: Among multiple radar modules, the current position and corresponding detection range of each radar module are determined based on the identification of each radar module, and the distribution position of multiple radar modules is determined according to the current position of each radar module; at the same time, the current position of the unmanned vehicle is collected, and the radar module sensing system is determined based on the distribution position of multiple radar modules, the corresponding detection range, and the current position of the unmanned vehicle.

[0012] In the embodiments of this application, the vehicle-mounted radar system performs self-tests and initialization, rather than sensing the external environment. The domain controller (DCU) periodically sends heartbeat packets or diagnostic commands to all radar nodes via CAN or vehicle Ethernet. This process mainly includes three levels: first, power status detection to confirm that the power supply to each radar module is stable; second, communication link detection to verify that the data path (such as FPD-Link) between each module is unobstructed; and third, by reading the internal registers of the radar chip, to confirm that its core processor and radio frequency front-end are working normally and without any fault codes. When the unmanned vehicle is powered on or the system is reset, this step is actively triggered to start the enumeration and recognition process of the entire radar system.

[0013] After confirming that all radar modules are online, the system immediately starts the data acquisition service and begins to receive the raw data streams they generate. The different paths here have a dual meaning: one refers to the physical spatial path, that is, the direction of propagation of radar electromagnetic waves in space, such as the forward detection path of the vehicle's front radar; the other refers to the data transmission path, that is, the data streams from different radar modules that are distinguished by different physical interfaces (such as different FPD-Link deserializer ports) or packet header identifiers at the domain controller end. At this stage, the system knows that several data streams are coming in, but has not yet associated them with specific physical radar entities.

[0014] The system uniquely binds data streams to physical modules by analyzing the fingerprint information they carry. There are three main traceability mechanisms: the most common is built-in ID traceability, where each radar module is programmed with a unique global identifier (UID) during production and periodically embedded in the data stream, which the domain controller can easily parse for identification; the second is physical port mapping traceability, where the mapping relationship between physical ports and installation locations is predefined during vehicle design, and the system can locate the module by directly looking up the port number of the incoming data stream; and the last is signal feature traceability, where in more complex scenarios, the system can instruct a specific radar to emit a special calibration signal, and the binding is completed by detecting which data stream exhibits this feature.

[0015] After tracing is completed, the system creates a logical instance for each identified radar module at the software level and assigns it a program-readable structured label. This label typically includes a logical name (such as FrontRadar), a physical ID (such as RADAR_A_UID_FRONT_001), a data stream handle (pointing to Stream_1), and status attributes (active, standby, fault, etc.). This labeling process is dynamic, allowing the system to handle radar hot-plugging or fault switching at runtime.

[0016] Specifically, at the moment the autonomous vehicle starts, its domain controller sends a broadcast signal to the vehicle's high-speed serial bus, which is equivalent to calling out a roll: all modules based on the satellite radar architecture, please report your status; after receiving the signal, each of the A radar modules distributed at the front, four corners and rear of the autonomous vehicle returns a health status packet, indicating that its power supply, communication and internal chips are all operating normally.

[0017] The domain controller of the autonomous vehicle will receive multiple high-speed serial data simultaneously; one data stream enters through physical port 1, another through port 2, and so on. These data streams are all 1-bit high-speed serial signals directly output by the Delta-Sigma modulator based on the satellite radar architecture. The system temporarily labels these data streams as Stream_1, Stream_2, Stream_3... It understands that these signals represent the detection results in different directions around the autonomous vehicle, but has not yet established a connection between Stream_1 and the physical entity of the front radar.

[0018] The domain controller begins parsing the data packets in Stream_1 and finds a unique ID at a specific location in the data packet: RADAR_A_UID_FRONT_001. The system immediately queries the pre-stored configuration table and confirms that this ID corresponds to the satellite radar module installed in the front grille of the autonomous vehicle. Similarly, it finds RADAR_A_UID_REAR_002 in Stream_2, which corresponds to the radar at the rear of the vehicle. Through this efficient ID tracing, the system successfully maps the abstract data stream to the specific physical radar module.

[0019] Through tracing, the autonomous vehicle's control system now has a clear list of radar modules: Radar_Front(logical marker)>Physical location: front grille>Data stream:Stream_1>UID:RADAR_A_UID_FRONT_001; Radar_Rear(logical marker)>Physical location: rear bumper>Data stream:Stream_2>UID:RADAR_A_UID_REAR_002.

[0020] Furthermore, the system obtains the precise installation coordinates (x, y, z) and installation attitude angles (yaw, pitch, roll) of each radar module in the vehicle coordinate system by reading the vehicle design parameters pre-stored in the non-volatile memory. At the same time, the system reads the performance parameters in the radar module firmware or configuration file, such as the maximum detection range, horizontal and vertical field of view, range and velocity resolution, etc. These parameters together define the spatial geometry that the radar can sense.

[0021] The system aggregates the position coordinates of all radar modules in the vehicle coordinate system obtained in the previous step. This is not only a list of coordinates, but also a structured description of spatial relationships. Through this layout, the system can calculate the sensor spacing to assess the blind zone size, determine the geometric intersection of different radar detection ranges to find the key area for data fusion, and generate a network topology that describes the spatial adjacency of sensors.

[0022] Acquiring high-precision pose information of an unmanned vehicle in the world coordinate system is usually accomplished by a dedicated positioning module. Its input sources include the Global Navigation Satellite System (GNSS), inertial measurement unit (IMU), wheel speedometer, and high-precision map. Through fusion algorithms such as extended Kalman filter or particle filter, the system can output a vehicle position that includes longitude, latitude, altitude, and heading angle.

[0023] The distribution locations of multiple radar modules, their corresponding detection ranges, and the current position of the autonomous vehicle are subjected to a complete coordinate transformation and spatial integration to generate a dynamic, vehicle-centric, and world-aligned unified perception model. This system includes a unified coordinate origin (the current position of the autonomous vehicle), all radar perception sectors that have been rotated and translated, and a comprehensive perception coverage polygon, which is refreshed in real time at a frequency of 10-100Hz.

[0024] Specifically, on the autonomous vehicle, the domain controller has already identified Radar_Front via S111. Now, it queries the configuration file and obtains that the installation position of Radar_Front is (3.5, 0.0, 0.5) meters in the vehicle coordinate system, that is, 3.5 meters in front of the vehicle and 0.5 meters above the center line. At the same time, it reads the performance parameters of the satellite radar module: maximum detection range of 250 meters and horizontal field of view of 120°. The system binds these attributes to the logical instance of Radar_Front. Similarly, it also binds the position and detection range parameters of modules such as Radar_RearLeft to their respective positions.

[0025] The autonomous vehicle's control system now has a set of satellite radar coordinates; through calculation, the system concludes that the autonomous vehicle adopts a four-corner radar layout scheme of front-rear-left-rear-right-rear; it further analyzes that the detection range of the left and right rear radars overlaps in a region directly behind the vehicle, and the data in this region can be provided by both radars simultaneously.

[0026] The driverless car is driving on a city road, and its positioning module is integrating data from GNSS, IMU and wheel speedometer. At this moment, it outputs the current position of the driverless car as: 116.4074° east longitude, 39.9042° north latitude, 50 meters above sea level, and heading angle of 90° (due east).

[0027] The autonomous vehicle's control system performed this final step; it transformed the sensing sectors of all satellite radars based on the vehicle's world coordinates and a heading angle of 90°; the 120° sector of Radar_Front, which was originally facing directly in front of the vehicle, was now rotated to face due east; the system generated a complete radar module sensing system, which is a dynamic sensing field covering a range of hundreds of meters around the autonomous vehicle; when a point cloud target detected by radar appears in this sensing field, the system can not only know the target's position relative to the vehicle, but also instantly calculate the target's precise latitude and longitude on the world map through coordinate inverse transformation.

[0028] refer to Figure 3 In step S12, the specific steps are as follows: S121: Real-time monitoring of radar module sensing system, determining multiple sensing contents based on the detection of radar module sensing system, and determining the sensing events of radar module according to each sensing content, the corresponding radar module and the working history of the radar module, so as to mark multiple sensing events; S122: Multiple radar modules interact and mark the interaction events between each radar module. At the same time, a satellite radar is constructed based on the distribution location, corresponding working data and corresponding functions of multiple radar modules. A control system for the satellite radar is constructed based on the radar architecture of the satellite radar, multiple sensing events and multiple radar modules.

[0029] In the embodiments of this application, the system continuously polls and subscribes to the radar module sensing system constructed in S11, and receives the raw data streams of all marked radar modules in real time through a high-speed bus. In the domain controller, the signal processing algorithm chain performs operations such as fast Fourier transform, constant false alarm rate detection, and cluster analysis on these raw signals, and finally parses the data stream into a structured target list containing physical attributes such as range, relative velocity, azimuth angle, and signal-to-noise ratio (SNR). At the same time, the system also monitors the environmental background status, such as clutter level and electromagnetic interference intensity.

[0030] The system interprets the target list extracted in the previous step using preset rules and a simple state machine. This includes: roughly classifying the targets based on their kinematic characteristics and physical properties (such as stationary objects, oncoming vehicles, and vehicles traveling in the same direction); describing the relative positional relationship between the target and the vehicle body (such as within the same lane, adjacent lane, and blind spot); and judging the target's movement trend based on several consecutive frames of data (such as approaching, moving away, or cutting in).

[0031] The system combines the sensed content with deeper contextual information for logical reasoning, ultimately triggering a sensed event with clear action guidance. The fused context includes: the sensed content itself, the function and reliability of the corresponding radar module, and a key historical experience parameter—the radar module's operational history. The system queries the radar's historical data performance in similar scenarios, such as whether false alarms frequently occur at this location, and whether the change in the target's radar cross-section (RCS) value matches the characteristics of a real vehicle. An evaluation is performed through a rule-based or machine learning model-based decision engine, and a sensed event is triggered when multiple conditions meet preset thresholds.

[0032] Each sensing event confirmed by the decision engine is encapsulated into a standardized, structured data object and placed in a global event queue for subsequent steps to call. A standard event object typically includes fields such as: a globally unique identifier (EventID), event type (EventType), precise timestamp (Timestamp), trigger source radar module (SourceModule), confidence level (ConfidenceLevel), priority (Priority), key parameters related to the event (Parameters), and event status (Status).

[0033] Specifically, the autonomous vehicle's domain controller is processing data streams from its four satellite radar modules at a frequency of 20 times per second. For Radar_Front, the system has just completed one processing cycle and output a list containing three targets: Target_1 (distance 80m, relative speed -15km / h, azimuth 0°), Target_2 (distance 150m, relative speed -5km / h, azimuth 2°), and Target_3 (distance 50m, relative speed +25km / h, azimuth -10°). At the same time, the system detects that the current SNR is stable at 25dB, indicating that the environmental interference is small. This processing is performed synchronously on all radar modules.

[0034] Based on the detection results from the previous step, the system generated multiple sensing data for the autonomous vehicle: for Target_1, the system determined its type to be Same Direction_Vehicle, its spatial relationship to be In_Lane, and its motion trend to be Approaching; for Target_3, the system determined its type to be Same Direction_Vehicle, its spatial relationship to be In_Adjacent_Lane (left side), and its motion trend to be Receding (overtaking). These sensing data are easier to understand than the original data, but they do not yet constitute an event that requires an immediate response.

[0035] The autonomous vehicle's system is processing the Radar_Front's sensing data regarding Target_1. The sensing data includes Same_Direction_Vehicle, In_Lane, and Approaching information. The source is Radar_Front, whose primary function is forward collision warning, and the data is highly reliable. The system queries the historical database and finds that this radar has an extremely low false alarm rate in this type of scenario, and the RCS value of Target_1 is stable, consistent with real vehicle characteristics. The decision engine calculates the time to collision (TTC) to be approximately 19.2 seconds. Although this does not reach the warning threshold, the system has already marked it as a high-priority tracking event. Event_ID: Evt_003, Type: Lead_Vehicle_Tracking, Source: Radar_Front, Confidence: 99%, Urgency: Medium; If the vehicle suddenly decelerates and the TTC is less than 3 seconds, the system will immediately trigger a Forward_Collision_Warning event. After the complete S121 process, the autonomous vehicle's event queue now contains multiple tagged sensor events, these clear and structured event objects.

[0036] Furthermore, multiple radar modules interact and the interaction events between each radar module are marked. At the same time, a satellite radar is constructed based on the distribution location, corresponding working data, and corresponding functions of multiple radar modules. A control system for the satellite radar is constructed based on the radar architecture of the satellite radar, multiple sensing events, and multiple radar modules. This system takes into account the overall consideration of the distribution location, corresponding working data, and corresponding functions of multiple radar modules, ensuring the accuracy of the satellite radar.

[0037] At this point, the system continuously compares target attributes (such as range, velocity, RCS) and timestamps from events from different radar modules through a spatiotemporal correlation engine. When a target moves from the field of view of one radar to the expected detection path of another radar within a very short time window, the system determines that it is the same target and stitches together and predicts its trajectory in different coordinate systems. When this correlation analysis meets a specific pattern, the system generates an interactive event, such as cooperative tracking, crossing / cutting in, or ghosting elimination, thereby describing the complete process that requires multiple radars to work together to detect the target.

[0038] Based on the distribution location and detection range of all radar modules, the system calculates the maximum sensing boundary of the entire radar system through geometric union operation; it establishes a standardized internal data model, converting all target information into a unified global coordinate and absolute velocity format; at the same time, the system aggregates the functions of each physical radar (such as ACC control and blind spot monitoring) to form a unified capability set and provides standardized service interfaces for upper-layer applications.

[0039] It integrates three core inputs: logical satellite radar providing a unified sensing field of view, all sensing / interaction events describing key dynamics, and radar architecture content including hardware implementation details (such as Delta-Sigma modulators, FPD-Link bus protocols, domain controller computing load, etc.). Based on these inputs, the management and control system can execute a series of advanced management strategies: when a high-priority event is detected, resources are dynamically scheduled, and relevant radars are instructed to increase their scanning frequency; the complexity of signal processing algorithms is dynamically adjusted according to computing power consumption to achieve adaptive performance optimization.

[0040] Specifically, the autonomous vehicle's system processes the Blind_Spot_Intrusion event (Evt_002) from Radar_RearLeft and the Cut_In_Warning event (Evt_004) from Radar_FrontLeft almost simultaneously. Spatiotemporal correlation engine analysis reveals that the targets in the two events are highly consistent in terms of relative speed and RCS value change trends, and the latter's location is exactly the expected location of the former as it travels along its current trajectory. Therefore, the system determines that these are the same target and marks an interaction event: Interaction_Event_ID:I_002,Type:Continuous_Tracking_Cut_In. This event provides a more complete description of the entire process of a vehicle rapidly cutting into the front lane from the left rear blind spot than a single event.

[0041] The autonomous vehicle's control system performs this step, constructing a logical entity called B_Vehicle_Satellite_Radar. It integrates the positions of the four A radars (front, rear, left rear, and right rear) and continuously monitors their operational data, such as temperature and bit error rate, forming a comprehensive system health status. Most importantly, its exposed capabilities include Get360DegreeView, GetLeadVehicleForACC, and GetBlindSpotTargets. This logic integrates the capabilities of all satellite radar modules, making it a powerful and comprehensive super radar for upper-layer applications.

[0042] The autonomous vehicle's satellite radar control system is fully activated upon receiving the I_002 (rapid entry) event. Based on the radar architecture, it determines that the domain controller has 30% of its computing power remaining and immediately makes a decision: to send instructions to Radar_FrontLeft and Radar_RearLeft, utilizing the high-speed sampling characteristics of the Delta-Sigma modulator to double the point cloud density of the target area to obtain more accurate trajectory information. Simultaneously, it transmits this high-priority event and its processing status via the API of the logical entity B_Vehicle_Satellite_Radar.

[0043] refer to Figure 4 In step S13, the specific steps are as follows: S131: Real-time monitoring of the autonomous driving moments of the autonomous vehicle, collection of multiple autonomous driving data points of the autonomous vehicle in autonomous driving state, and determination of the autonomous driving mode of the autonomous vehicle based on multiple autonomous driving data points, the driving behavior of the autonomous vehicle and the corresponding current location; S132: Based on the identification of the satellite radar control system, determine multiple surrounding environmental features of the unmanned vehicle, and determine the dynamic control content of the unmanned vehicle based on the multiple surrounding environmental features, the unmanned driving mode of the unmanned vehicle, and the working status of the satellite radar itself. S133: Collect the surrounding environment detected by the unmanned vehicle, determine the first level of intelligent control content based on the surrounding environment detected by the unmanned vehicle and the dynamic control content of the unmanned vehicle, determine the second level of intelligent control content based on the surrounding environment detected by the unmanned vehicle and the working history of the satellite radar, and determine the intelligent control event of the unmanned vehicle based on the first level of intelligent control content and the second level of intelligent control content.

[0044] In the embodiments of this application, the autonomous driving time of the unmanned vehicle is monitored in real time, and multiple autonomous driving data of the unmanned vehicle in the autonomous driving state are collected. Based on the multiple autonomous driving data, the driving behavior of the unmanned vehicle and the corresponding current position, the autonomous driving mode of the unmanned vehicle is determined. This approach takes into account the overall consideration of multiple autonomous driving data, the driving behavior of the unmanned vehicle and the corresponding current position, thus ensuring the accuracy of the autonomous driving mode of the unmanned vehicle.

[0045] At this time, the system subscribes to and caches data streams directly related to driving behavior and vehicle status through the vehicle network (such as CAN-FD or vehicle Ethernet). This includes three types of core data: first, operation command data, such as the target trajectory, target steering wheel angle, and target acceleration issued by the planning module; second, vehicle status data, such as the actual vehicle speed, steering wheel angle, yaw rate, and actual lane position fed back by the actuator; and third, system activation status data, such as whether ADAS functions such as ACC / LKA are activated, whether the navigation system is guiding, and whether the HMI has driver intervention.

[0046] The system utilizes behavior recognition models (such as rule-based state machines or classification models) to identify driving behavior. Longitudinal behavior analysis identifies behaviors such as cruising, following, acceleration, and deceleration by analyzing the difference between the target and actual acceleration and changes in following distance. Lateral behavior analysis identifies behaviors such as lane keeping, lane change preparation, and lane change execution by analyzing the rate of change of steering wheel angle and the vehicle's position within the lane. The system deeply integrates the identified behaviors with high-precision maps and positioning information, matching current road attributes (such as highways, urban roads, curves, and ramps) and traffic environment (such as traffic density and speed limits). The decision engine integrates all the above information and, through decision trees or logical reasoning, ultimately determines the most suitable and accurate autonomous driving mode for the current situation, such as Highway_Cruising, Urban_Cut-In, or Emergency_Stop.

[0047] Specifically, the driverless car is traveling at 100 km / h in the middle lane of the highway. The system is collecting the following driverless data in real time: In terms of operation commands, the target path issued by the planning module is a straight line, the target speed is 100 km / h, and the target acceleration is 0. In terms of vehicle status, the actual speed is stable at 100 km / h, the steering wheel angle is 0.1° (fine-tuning), and the vehicle is located in the center of the lane. In terms of system status, the ACC status is Active, the LKA status is Active, the navigation system displays no active navigation commands, and the HMI shows no driver intervention.

[0048] The autonomous vehicle's system processes the collected data: In terms of behavior recognition, longitudinal analysis shows stable vehicle speed and a long following distance, identifying it as Cruising; lateral analysis shows minimal steering wheel angle and the vehicle centered, identifying it as Lane_Keeping; in terms of location association, the high-precision map shows the current location as G2 Beijing-Shanghai Expressway, K150+200 section, a straight line with a speed limit of 120km / h; the decision engine integrates the behaviors of Cruising and Lane_Keeping, as well as the location attributes of the Highway, and finally determines that the current autonomous driving mode of the autonomous vehicle is Highway_Cruising (highway cruise mode). The determination of this mode means that the subsequent control logic will prioritize the smoothness and stability of driving, and adopt stricter triggering conditions for high-risk operations such as lane changes.

[0049] Furthermore, based on the identification of the satellite radar control system, multiple surrounding environmental features of the unmanned vehicle are determined. Based on these multiple surrounding environmental features, the unmanned driving mode of the unmanned vehicle, and the working status of the satellite radar, the dynamic control content of the unmanned vehicle is determined. This comprehensive consideration of multiple surrounding environmental features, the unmanned driving mode of the unmanned vehicle, and the working status of the satellite radar ensures the accuracy of the dynamic control content of the unmanned vehicle.

[0050] At this point, the system requests a snapshot of the current environment from the logical satellite radar entity constructed by S12 via API or internal message bus, and parses out a series of key environmental features from it. These features are usually semantic key-value pairs or structured objects, including: a detailed target list containing all tracked target types, locations, and speeds; lane information such as clarity, width, and curvature of the current lane and adjacent lanes; and comprehensive indicators reflecting spatial relationships and environmental situations, such as distance to the vehicle in front, collision time, blind spot occupancy status, and cutting-in risk index.

[0051] Based on the driving mode determined by S131 (e.g., Highway_Cruising), the system loads the corresponding control strategy parameter set, which defines the core objectives of the mode (e.g., prioritizing smoothness). The decision engine matches environmental features with the strategy parameter set for calculation, such as determining longitudinal control based on the distance to the vehicle ahead and TTC, and assessing lane change feasibility based on the status of adjacent lanes. Simultaneously, the system performs risk assessment and generates constraints such as No_Lane_Change. Furthermore, the radar's own operating status also affects the decision; for example, a decline in the performance of a radar may cause the system to reduce its data weight or require other sensors to enhance verification. The system outputs a set of structured control commands and parameters, defining the vehicle's specific behavior in the next few seconds.

[0052] Specifically, the autonomous vehicle's control system queried its satellite radar management system and obtained the following surrounding environmental characteristics: Lead_Vehicle_Distance: 85m; Lead_Vehicle_Relative_Speed: -5km / h (the vehicle in front is slower than vehicle B); Time_To_Collision: 61.2s; Left_Lane_Status: Clear (no vehicle in the left lane); Right_Lane_Target: Present, Distance: 120m, Relative_Speed: +20km / h (a vehicle is rapidly approaching from the right rear); Blind_Spot_Status: All_Clear; Cut_In_Risk_Index: Low (the overall assessment indicates low risk of cutting in).

[0053] The autonomous vehicle's system executes fusion decision-making: the mode is Highway_Cruising, and the goal is smooth cruising; environmental characteristics show that the vehicle ahead is far away, the left lane is empty, and there is a vehicle on the right; all A radar modules are functioning normally; based on this, the system decides: in longitudinal control, there is no threat from the vehicle ahead, so maintain the set speed of 100km / h; in lateral control, although the left lane is clear, there is currently no intention to actively change lanes, so keep the lane centered; at the same time, due to a rapidly approaching vehicle on the right, a constraint prohibiting lane changes to the right is generated; the system determines the dynamic control content of the autonomous vehicle as follows: Longitudinal_Control_Mode:Speed_Control;Target_Speed:100km / h;Lateral_Control_Mode:Lane_Keeping;Target_Lane:Current_Lane;Maneuver_Constraints:{Prohibit_Right_Lane_Change:True}.

[0054] Therefore, the system collects data on the surrounding environment detected by the autonomous vehicle (V2V). Based on this data and the V2V's dynamic control behavior, a first level of intelligent control is determined. A second level of intelligent control is determined based on the same data and the operational history of the satellite radar. The intelligent control events of the V2V are then determined based on both levels of intelligent control, ensuring the accuracy of the intelligent control events. Furthermore, a radar module sensing system is introduced to manage the satellite radar system, further enhancing the accuracy of the intelligent control events by considering the V2V's dynamic control behavior, the detected surrounding environment, and the operational history of the satellite radar.

[0055] At this point, the system will simultaneously collect data from other sensors such as the forward-looking camera and LiDAR, as well as high-precision maps, and compare the environmental features perceived by the satellite radar with these broader perception results; for example, radar-camera fusion can be used to enhance the reliability of target recognition, or radar-map fusion can be used to confirm static obstacles; if a conflict is detected between different sensors (such as the radar thinking there is no vehicle in front, but the camera detecting a stationary object), the system will adopt a more conservative strategy; based on the fused environmental information, the system confirms the dynamic control content of S132 and generates a first layer of intelligent control content with high confidence that has been verified by multiple sensors.

[0056] The system uses the current surrounding environmental features as query conditions to search through a massive historical database and uses machine learning models to find historical cases similar to the current environment. The purpose is to discover potential, non-explicit long-tail risks, such as ghost targets generated by radar under specific road and weather combinations, or learn drivers' following habits under specific traffic flows. Based on the historical patterns and risks discovered, the system generates a set of optimization suggestions or enhanced constraints, forming the second layer of intelligent control.

[0057] The system employs an arbitration mechanism to handle the relationship between the two layers of content. Typically, the first layer of content (real-time verification) has a higher priority because it reflects the most realistic physical world at present, while the second layer of content (historical prediction) serves as a supplement and optimization, fine-tuning the control parameters without conflict. The system encapsulates the final decision after fusion into a standardized intelligent control event.

[0058] Specifically, when the autonomous vehicle performs this step, its dynamic control involves maintaining a speed of 100 km / h and keeping the vehicle centered in its lane. The system collects a wider range of environmental data: the forward-facing camera confirms all vehicles detected by the radar, and the lane lines are clear and without anomalies; the high-precision map shows a straight road 2 kilometers ahead with no entrances or exits; all sensor data are highly consistent, confirming the low-risk state of the current environment; therefore, the first level of intelligent control determined by the system is: Maintain_Speed_100kph,Lane_Keep_Center.

[0059] When the autonomous vehicle performs this step, the system queries the database and finds that under the current road conditions and similar weather conditions, when a vehicle approaches from the right lane at a relative speed of +20 km / h, there is a 3% probability that the vehicle will suddenly change lanes to the left within the next 5 seconds. The system identifies this as a potential right-side lane-cutting risk, even though the current radar's Cut_In_Risk_Index is Low. Therefore, the system determines the second level of intelligent control as follows: Add_Constraint:{Increase_Right_Side_Margin:True},Add_Monitor:{Enhance_Right_Radar_Tracking:True} suggests that the system actively maintain a greater lateral distance from the vehicle on the right and allocate more computing power to accurately track the target.

[0060] The autonomous vehicle's system executes the final decision: the first set of parameters is Maintain_Speed_100kph and Lane_Keep_Center; the second set is Increase_Right_Side_Margin and Enhance_Right_Radar_Tracking. There is no conflict between the two sets of parameters; the second set is a refinement and enhancement of the first. The system decides to adopt all suggestions and ultimately generates an intelligent control event. Event_ID:Ctrl_Evt_002,Type:Highway_Cruising_with_Cautious_Monitoring,Parameters:{Longitudinal:{Target_Speed:100kph},Lateral:{Mode:Lane_Keep,Target_Offset:-0.1m (slightly to the left to increase right clearance)},Monitoring:{Right_Radar_Priority:High}}.

[0061] refer to Figure 5 In step S14, the specific steps are as follows: S141: Real-time monitoring of intelligent control events of unmanned vehicles, identification of intelligent control events of unmanned vehicles, output of multiple item tags, determination of corresponding intelligent control items based on the tracing of each item tag, and collection of multiple intelligent control items. S142: Based on the identification of each intelligent control item, determine the item content of each intelligent control item, and determine the item priority of each intelligent control item according to the comparison of each intelligent control item; at the same time, collect the driving data of the unmanned vehicle; S143: Determine the first intelligent control coefficient based on the driving data of the unmanned vehicle and the project content of each intelligent control project; determine the second intelligent control coefficient based on the driving data of the unmanned vehicle and the project priority of each intelligent control project; determine the intelligent control state of the unmanned vehicle according to the mapping relationship between the first intelligent control coefficient, the second intelligent control coefficient and the intelligent control state.

[0062] In the embodiments of this application, the system subscribes to a specific topic through a highly reliable internal communication mechanism (such as DDS or SOME / IP), and the S13 module publishes all generated intelligent control events to the topic; when an event message is received, the system parses its message header, obtains the event's metadata, and most importantly, parses the EventType field. This predefined enumeration value (such as Highway_Cruising, Emergency_Braking) allows the system to immediately identify the intent and scope of the macro decision.

[0063] The system maintains a statically or dynamically configured event-to-project mapping table, which defines which basic, atomic control projects are needed to complete each type of macro-level intelligent control event. The system queries the mapping table based on the EventType identified in the previous step, and the mapping table returns one or more project markers. These markers are the keys or indexes for subsequent searches of specific project content.

[0064] The system uses the project tags obtained in the previous step to trace and extract the specific control commands and parameters associated with them; the system returns to the original event message published in S13 and looks up the specific parameter values ​​corresponding to each project tag in its Parameters field; the system combines the project tags and the traced parameters together to create a complete, structured intelligent control project object.

[0065] The system aggregates all the smart control project objects generated in the previous step into an ordered list or set, and passes it as the final output of module S141 to the next sub-step S142; the system creates a container (such as a list) and puts all the instantiated smart control projects into it in a certain order (such as by priority or functional module).

[0066] Specifically, the system subscribes to a specific topic through a highly reliable internal communication mechanism (such as DDS or SOME / IP). The S13 module publishes all generated intelligent control events to this topic. When an event message is received, the system parses its message header to obtain the event's metadata, most importantly parsing the EventType field. This predefined enumeration value (such as Highway_Cruising, Emergency_Braking) allows the system to immediately identify the intent and scope of the macro-level decision. The autonomous vehicle's system uses EventType:Highway_Cruising_with_Cautious_Monitoring to query its internal mapping table; the mapping table returns the following three item tags: Proj_Marker_1:Longitudinal_Speed_Control; Proj_Marker_2:Lateral_Lane_Keeping; Proj_Marker_3:Enhanced_Monitoring, these three markers represent the three core task modules necessary to achieve the macro-level intent of cautious cruise.

[0067] The autonomous vehicle's system uses three item tags to trace the parameters of the Ctrl_Evt_002 event: for the tag Longitudinal_Speed_Control, the system finds {Target_Speed:100kph} in the event parameters; for the tag Lateral_Lane_Keeping, the system finds {Mode:Lane_Keep,Target_Offset:-0.1m}; and for the tag Enhanced_Monitoring, the system finds {Focus_Area:Right_Side,Priority:High}. Through tracing, the system identifies three complete intelligent control items, each containing a clearly defined type and execution parameters.

[0068] The autonomous vehicle's system constructs the three intelligent control projects mentioned above into a list called Active_Control_Projects. This list now contains all the specific control tasks that the autonomous vehicle needs to perform at the current moment. This list is passed in its entirety to the S142 module, which will determine their priorities based on the content of these projects and, in conjunction with the vehicle's driving data, ultimately calculate the intelligent control state of the autonomous vehicle.

[0069] Furthermore, the content of each intelligent control project is determined based on the identification of each intelligent control project, and the priority of each intelligent control project is determined based on the comparison of each intelligent control project. At the same time, the driving data of the unmanned vehicle is collected, which takes into account the overall consideration of the comparison of each intelligent control project and ensures the accuracy of the priority of each intelligent control project.

[0070] At this point, the system iterates through each intelligent control project input from S141, extracting its core attributes. The intelligent control project includes the project type and original parameters. The system converts the original parameters into standard units and formats used within the control system (e.g., converting 100kph to 27.78m / s) and performs a rationality check to ensure that the parameters are within the physically permissible and safety limits. After parsing and verification, the project content of each intelligent control project is finally determined as a structured and standardized instruction set.

[0071] The system employs a weighted scoring model, with evaluation dimensions including: the inherent safety level defined according to functional safety standards (such as ISO 26262); the situational urgency dynamically assessed based on the current environment; the functional dependencies between different items; and the detected driver intent. The system uses this model to score all items in parallel, obtaining a quantified priority score (such as 0-100), and sorts all items according to the score to form a clear execution order.

[0072] The system collects real-time vehicle status data at a high frequency (typically 50-200Hz) via the vehicle's CAN / CAN-FD bus or in-vehicle Ethernet. This includes longitudinal status (current vehicle speed, acceleration), lateral status (steering wheel angle, yaw rate, real-time lateral position of the vehicle within the lane), and other statuses (gear, torque output, etc.). The collected data is precisely timestamped and time-aligned with the evaluation results of the control items.

[0073] Specifically, the autonomous vehicle's system receives three intelligent control items from S141 and determines their content: The original parameters {Target_Speed:100kph} of Project 1 (Longitudinal_Speed_Control) were normalized to {Target_Speed:27.78m / s,Control_Mode:PID_Speed}. After confirming that the offset of the original parameter {Mode:Lane_Keep,Target_Offset:-0.1m} for Project 2 (Lateral_Lane_Keeping) was reasonable, the content was determined to be {Control_Mode:Stanley_Lateral,Lateral_Offset:-0.1m}. The content of Project 3 (Enhanced_Monitoring) is determined to be {Monitoring_Task:Blind_Spot_And_Cut_In,Sensor_Allocation:Right_Radar_High_Priority}.

[0074] The autonomous vehicle system prioritized three items: Item 1 (speed control) and Item 2 (lane keeping) were routine cruise tasks with an inherent safety level of QM and low situational urgency, both scoring 40; Item 3 (enhanced monitoring), although not directly controlling the vehicle, had a parameter Priority: High indicating that it was associated with a potential collision risk (right-side entry), and according to the situational urgency model, the score for this item was significantly increased to 85; after comparison, the system determined the priority of each item: Item 3 (highest) > Item 1 = Item 2 (tied, medium).

[0075] While evaluating project priorities, the autonomous vehicle's system collected the following driving data: Current_Speed: 27.72m / s (99.8km / h); Longitudinal_Acceleration: 0.05m / s²; Steering_Wheel_Angle: 1.5°; Lateral_Offset: -0.05m (vehicle centerline is 5cm to the left of lane centerline); Yaw_Rate: 0.1° / s. This data, along with the project content and priorities determined in the first two steps of S142, will be passed to S143 to finally calculate the current intelligent control state of the autonomous vehicle.

[0076] Therefore, a first intelligent control coefficient is determined based on the autonomous vehicle's driving data and the content of each intelligent control project; a second intelligent control coefficient is determined based on the autonomous vehicle's driving data and the project priority of each intelligent control project; and the intelligent control state of the autonomous vehicle is determined according to the mapping relationship between the first intelligent control coefficient, the second intelligent control coefficient, and the intelligent control state. This approach takes into account the overall consideration of the mapping relationship between the first intelligent control coefficient, the second intelligent control coefficient, and the intelligent control state, ensuring the accuracy of the autonomous vehicle's intelligent control state.

[0077] At this point, the system establishes an error calculation channel for each intelligent control item. For example, it calculates Speed_Error for speed control and Lateral_Error for lane keeping. Since different physical quantities have different dimensions, the system uses a normalization function (such as the Sigmoid function) to transform each error value into a unified, dimensionless range (such as 0 to 1, where 0 represents a perfect match). The system assigns weights to the corresponding normalized errors according to the importance of each control item and merges all errors into a single total error value through weighted averaging. The system uses a monotonically decreasing function (such as f(x) = 1 - x) to convert the total error into the first intelligent control coefficient. The closer the value is to 1, the higher the degree of matching between the vehicle's current state and the expected command.

[0078] The system maps the project priorities (e.g., Critical, High, Medium, Low) determined in S142 to numerical values ​​(e.g., Critical=1.0, High=0.8). The system will dynamically adjust the priorities based on driving data, for example, the value of the Medium task will be slightly increased under high-speed conditions. During fusion, the system tends to use the maximum value or the softmax function to ensure that a high-priority task can significantly increase the urgency assessment of the entire system. The fused result is the second intelligent control coefficient, and the closer its value is to 1, the higher the overall urgency or importance of the current control task.

[0079] The system defines a finite set of intelligent control states, such as Stable_Cruising, Cautious_Cruising, and Active_Intervention. The system establishes a mapping model from two-dimensional inputs (first coefficient, second coefficient) to one-dimensional outputs (control states). This model can be a rule-based two-dimensional lookup table or a classifier (such as a decision tree). The system feeds the two calculated coefficients as inputs into the mapping model, and the output of the model is the current final intelligent control state of the autonomous vehicle.

[0080] Specifically, the autonomous vehicle system performs the following calculations: calculates the speed error |27.72-27.78|=0.06m / s and the lateral error |-0.05-(-0.10)|=0.05m. These errors are within acceptable limits and, after normalization, are 0.1 and 0.2 respectively. Assuming equal longitudinal and lateral control weights, the total error is (0.1+0.2) / 2=0.15. The first intelligent control coefficient = 1-0.15=0.85, which indicates that the current driving state of the autonomous vehicle is highly consistent with the expected command.

[0081] The autonomous vehicle's system performs this calculation: quantifies the priority of the tasks, mapping tasks 1 and 2 (Medium) to 0.5 and task 3 (High) to 0.8; since the current vehicle speed is 100km / h, the system fine-tunes the value of the Medium task to 0.55; since there is a High priority task, the system uses a fusion algorithm that highlights high priority, resulting in a comprehensive urgency score of 0.75; therefore, the second intelligent control coefficient is 0.75, indicating that although the vehicle is driving smoothly, there is currently a control task that requires high vigilance.

[0082] The autonomous vehicle's system makes the final decision: its input coefficients are a first coefficient of 0.85 (high matching degree) and a second coefficient of 0.75 (high urgency). The system searches for the coordinate point (0.85, 0.75) in its two-dimensional lookup table. This point falls within the Cautious_Cruising region. The logic of this region is that although the vehicle is precisely executing instructions, the overall situation is cautious due to the existence of a high-priority monitoring task. Therefore, the system determines the current intelligent control state of the autonomous vehicle to be Cautious_Cruising. This state accurately summarizes the core characteristics of the autonomous vehicle at this moment: precise control, but vigilance.

[0083] refer to Figure 6 In step S15, the specific steps are as follows: S151: Mark the current position of the autonomous vehicle, determine multiple driving factors and environmental factors based on the surrounding detection of the autonomous vehicle, and determine the driving scenario in which the autonomous vehicle is located based on the current position of the autonomous vehicle, multiple driving factors and environmental factors; S152: Collect the intelligent control status of the unmanned vehicle, determine the first level of instruction content based on the intelligent control status of the unmanned vehicle and the driving scenario in which the unmanned vehicle is located, and at the same time, collect the previous control events of the unmanned vehicle, determine the second level of instruction content based on the intelligent control status of the unmanned vehicle and the previous control events of the unmanned vehicle, and determine the current control instruction of the unmanned vehicle based on the first level of instruction content and the second level of instruction content. S153: Collect environmental data detected by satellite radar, determine corresponding obstacle data based on the dynamic detection of satellite radar, determine the first level of driving content based on the current control command of the unmanned vehicle and the environmental data detected by satellite radar, determine the second level of driving content based on the current control command of the unmanned vehicle and the obstacle data detected by satellite radar, and determine the corresponding intelligent driving content based on the first level of driving content and the second level of driving content.

[0084] In the embodiments of this application, the current position of the unmanned vehicle is marked, and multiple driving factors and environmental factors are determined based on the surrounding detection of the unmanned vehicle. The driving scenario in which the unmanned vehicle is located is determined according to the current position of the unmanned vehicle, multiple driving factors and environmental factors, which is compatible with the overall consideration of the current position of the unmanned vehicle, multiple driving factors and environmental factors, and ensures the accuracy of the driving scenario in which the unmanned vehicle is located.

[0085] At this point, the system employs a tightly coupled fusion positioning algorithm that continuously fuses global absolute coordinates from GNSS, high-frequency pose change data from IMU, precise mileage information from wheel speedometers, and the results of matching LiDAR or cameras with environmental features (such as lane lines and road signs) to correct accumulated errors. The fusion algorithm ultimately outputs a six-degree-of-freedom (6-DoF) vehicle pose in a unified coordinate system, including timestamps, three-dimensional coordinates, and three-dimensional attitude. This precise pose is the current marked position.

[0086] The system utilizes full-stack perception capabilities, including satellite radar, cameras, lidar, and high-precision maps. Driving factors mainly describe the static attributes of the road itself, such as road curvature, slope, number of lanes, lane width, and speed limit extracted from high-precision maps. Environmental factors mainly describe dynamic and meteorological conditions, such as weather conditions, visibility, and lighting conditions determined by sensors or image analysis, as well as traffic flow density, average vehicle speed, and road surface conditions statistically obtained through radar and cameras.

[0087] The system employs a rule-based or machine learning-based scene classification model, which receives all extracted factors as input. The rule-based model contains a series of IF-THEN logics; for example, if the road type is highway and the weather is sunny, the scene is normal highway cruising. The machine learning-based model learns the probabilistic relationship between different combinations of factors and specific scenes through a large amount of labeled data. The final output of the model is a predefined scene label, which is a key input for subsequent decision-making steps because it is directly related to the control strategies and parameter sets under different scenes.

[0088] Specifically, the unmanned vehicle is traveling on the G2 highway; its fusion positioning module integrates GNSS signals, IMU data, and LiDAR scans of road markings and guardrails to ultimately mark its current location as: Easting: 385123.45m, Northing: 4123456.78m, Heading: 89.5°, Lane: 2 (accurate to the second lane, heading angle is due east).

[0089] The autonomous vehicle's system, based on its surrounding detection, determined the following factors: in terms of driving factors, the road curvature was 0.001 / km (approximately a straight line), the slope was 0%, the number of lanes was 3, the current lane width was 3.75m, and the speed limit was 120km / h; in terms of environmental factors, the weather was sunny, the visibility was >1000m, the lighting was daylight, and the traffic flow density was medium (approximately 15 vehicles / km).

[0090] The autonomous vehicle's system inputs all factors (location = highway, number of lanes = 3, speed limit = 120km / h, weather = sunny, traffic density = moderate, etc.) into its scene classifier; the classifier matches the most suitable pattern and finally determines that the autonomous vehicle's current driving scenario is Highway_Cruising_Normal_Conditions. This scenario label tells the subsequent decision-making module that the current situation is a standard highway cruising condition, and appropriate control strategies that focus on comfort and efficiency should be adopted.

[0091] Furthermore, the intelligent control status of the autonomous vehicle is collected, and the first level of instruction content is determined based on the intelligent control status and the driving scenario in which the autonomous vehicle is located. At the same time, the previous control events of the autonomous vehicle are collected, and the second level of instruction content is determined based on the intelligent control status and the previous control events of the autonomous vehicle. The current control instruction of the autonomous vehicle is determined based on the first level of instruction content and the second level of instruction content. This overall consideration of the first level of instruction content and the second level of instruction content ensures the accuracy of the current control instruction of the autonomous vehicle.

[0092] At this point, the system takes the intelligent control state (such as Cautious_Cruising) output by S14 and the driving scenario (such as Highway_Cruising_Normal_Conditions) output by S151 as input and sends them to a policy mapping engine. The engine internally stores a multi-dimensional state-scenario-policy mapping table or rule base, which defines the macro-behavioral patterns that the vehicle should adopt under different combinations. Based on the matched policy, the engine generates a set of structured, descriptive rather than numerical instructions.

[0093] The system collects all control event sequences from a short-term cache over the past N seconds and analyzes this historical sequence using a pattern recognition or time series analysis model. The goal is to identify the trends, stability, and triggering sources of the current state of vehicle control behavior. Based on the analysis results, the system generates a set of instructions for smoothing and optimization.

[0094] The system uses an arbitrator to handle the relationship between the two sets of instructions. Typically, the first set of instructions (strategy) has higher priority and defines the control target and direction, while the second set of instructions (optimization) serves as a supplement, constraining and refining the execution process. The arbitrator transforms the fused instructions into an instruction set containing specific control parameters by calling the vehicle dynamics model, and finally outputs a standardized control instruction object.

[0095] Specifically, the autonomous vehicle's system executes this step with the intelligent control state (Cautious_Cruising) and the driving scenario (Highway_Cruising_Normal_Conditions) as inputs. The policy mapping engine matches the policy corresponding to this combination to perform a targeted lateral avoidance maneuver while maintaining high-speed cruising efficiency to address potential risks. Therefore, the first set of instructions determined by the system is: {Core_Task:Maintain_Highway_Cruise,Sub_Task:Execute_Lateral_Offset_For_Risk_Mitigation,Constraint:Maintain_Speed}.

[0096] The autonomous vehicle's system executes this step, querying the control history of the past 15 seconds. It finds that due to the continuous approach of the vehicle on the right, the system has issued three small left turn commands in the past 10 seconds, and the vehicle has already shifted approximately 5cm to the left from the lane centerline. The model recognizes this as a continuous, gradual lateral adjustment to avoid risk. Therefore, the system determines the second instruction as follows: {Optimization_Goal:Ensure_Control_Smoothness,Action:Continue_Gradual_Left_Offset,Target_Offset_Refinement:-0.1m}.

[0097] The autonomous vehicle's system executes the final decision: the first instruction requires lateral avoidance, and the second instruction requires smooth continuation; the arbitrator decides to adopt the smooth execution suggestion of the second instruction and combines it with the goal of the first instruction; it concretizes the descriptive task Execute_Lateral_Offset_For_Risk_Mitigation into a parameterized instruction; the system determines the current control instruction for the autonomous vehicle as: {Control_Mode:Lane_Keeping_with_Offset,Target_Speed:100km / h,Target_Lateral_Offset:-0.1m,Transition_Profile:Smooth}.

[0098] Therefore, by collecting environmental data detected by satellite radar and determining corresponding obstacle data based on the dynamic detection of satellite radar, the first level of driving content is determined according to the current control command of the unmanned vehicle and the environmental data detected by satellite radar. The second level of driving content is determined according to the current control command of the unmanned vehicle and the obstacle data detected by satellite radar. The corresponding intelligent driving content is determined based on the first and second driving content. This approach integrates the overall consideration of the first and second driving content, ensuring the accuracy of the corresponding intelligent driving content. At the same time, the intelligent control state of the unmanned vehicle is introduced to further control the current control command of the unmanned vehicle. This approach realizes the overall consideration of the current control command of the unmanned vehicle, the environmental data detected by satellite radar, and the obstacle data, thereby improving the accuracy of the intelligent driving content.

[0099] At this point, the system acquires structured environmental data, including target list, target status, static information, and quality information, from the satellite radar control system at high frequency. A risk assessment algorithm filters out targets that pose a potential threat to the current driving of the autonomous vehicle from all targets through spatial filtering, kinematic prediction, and time-of-collision (TTC) calculation. When the TTC of a target is less than the safety threshold or its predicted trajectory intrudes into the vehicle's safety envelope, the target is marked as an obstacle, and all its status information is included in the obstacle data subset.

[0100] The system analyzes the current control command of S152 and uses environmental data (including all targets, not just obstacles) as the world model for planning. The path planner generates a desired geometric path (such as a lane centerline with offset) according to the command, and the speed planner generates a desired speed curve on the path according to the target speed and the road conditions ahead. This spatiotemporal trajectory together constitutes the first layer of driving content.

[0101] The system uses a collision detector to compare the predicted trajectory of the first-level trajectory with the predicted trajectory of each obstacle, and evaluates the minimum distance and TTC. If the safety check fails or there is a high risk, the trajectory optimization module (such as the model prediction controller MPC) will intervene to fine-tune or replan the first-level trajectory while meeting the safety constraints. The final trajectory after safety check and correction is the second-level driving content.

[0102] The system selects the final trajectory through a simple arbitration logic: if a second layer of driving content exists (meaning a safety correction has been made), it is used; otherwise, the first layer is adopted. The system takes the target point of the next control cycle on the final trajectory as input and calculates the underlying control quantities required to achieve this target through the vehicle's inverse kinematics model. These include the target steering wheel angle and the target longitudinal acceleration / deceleration. This set of numerical instructions is encapsulated into CAN messages and sent to controllers such as EPS and ESP, thereby achieving precise and safe physical control of the driverless car.

[0103] Specifically, the autonomous vehicle's system acquired environmental data containing 15 targets from satellite radar. After analysis by the risk assessment algorithm, the following obstacle data was identified: Obstacle 1 (IDObj_08), located 120 meters to the right front, approaching at a relative speed of +20 km / h, with a TTC of 6.1 seconds, was marked as a high-concern obstacle due to its rapid approach, even though it just exceeded the threshold; Obstacle 2 (IDObj_12), located 200 meters directly in front, traveling at a relative speed of -5 km / h, was marked as a regular following target; other targets were ignored due to their excessive distance or lack of collision risk.

[0104] The system analyzes the current control command of S152 and uses environmental data (including all targets, not just obstacles) as the world model for planning. The path planner generates a desired geometric path (such as a lane centerline with offset) according to the command, and the speed planner generates a desired speed curve on the path according to the target speed and the road conditions ahead. This spatiotemporal trajectory together constitutes the first layer of driving content.

[0105] The autonomous vehicle's system performs this step: it performs collision detection between the first trajectory and obstacle data (especially Obj_08); the MPC controller analysis finds that performing an offset of -0.1m would slightly shorten the lateral safety distance to the rapidly approaching vehicle on the right; in order to maximize safety redundancy, the MPC decides to sacrifice some offset and replan a new trajectory that is only offset to the left by 0.05 meters, thus maintaining the avoidance intention while leaving more safety margin for the right side. This new trajectory, which has been optimized for safety, constitutes the second layer of driving content.

[0106] The autonomous vehicle's system performs the final calculation: it selects a second, safety-optimized trajectory; based on the target point on this trajectory in the next 100ms, the inverse kinematics model calculates that to achieve a deviation of -0.05m, the steering wheel needs to be turned 1.2° to the left; to maintain a speed of 100km / h, a small acceleration of 0.05m / s² is required; the system determines the intelligent driving content as: {Target_Steering_Angle:1.2°,Target_Longitudinal_Acceleration:0.05m / s²}, this set of instructions is sent to the chassis, and the autonomous vehicle will smoothly make a slight adjustment to the left and continue cruising at a stable speed.

[0107] Please see Figure 7 , Figure 7 This is a schematic diagram of the structural composition of the intelligent control system for an unmanned vehicle based on satellite radar according to an embodiment of the present invention; the intelligent control system for the unmanned vehicle based on satellite radar includes: The radar module sensing system module 21 is used to mark multiple radar modules based on the detection of the unmanned vehicle, and to determine the radar module sensing system according to the distribution location of the multiple radar modules, the corresponding detection range and the current position of the unmanned vehicle. The control system module 22 is used to determine multiple sensing events based on the identification of the radar module sensing system, and to construct a control system for the satellite radar based on each sensing event, the working data of each radar module, and the interaction events between each radar module. The intelligent control event module 23 is used to determine the dynamic control content of the unmanned vehicle based on the satellite radar control system and the unmanned driving mode of the unmanned vehicle, and to determine the intelligent control events of the unmanned vehicle based on the dynamic control content of the unmanned vehicle, the surrounding environment detected by the unmanned vehicle, and the working history of the satellite radar. The intelligent control status module 24 is used to determine multiple intelligent control items based on the identification of intelligent control events of the unmanned vehicle in the intelligent control events of the unmanned vehicle, and to determine the intelligent control status of the unmanned vehicle according to the item content, corresponding item priority and driving data of the unmanned vehicle. The intelligent driving content module 25 is used to collect the driving scene where the unmanned vehicle is located, determine the current control command of the unmanned vehicle based on the intelligent control status of the unmanned vehicle, the driving scene where the unmanned vehicle is located and the previous control events of the unmanned vehicle, and determine the corresponding intelligent driving content based on the current control command of the unmanned vehicle, environmental data and obstacle data detected by satellite radar.

[0108] The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

Claims

1. An intelligent control method for unmanned vehicles based on satellite radar, characterized in that, include: Multiple radar modules are marked based on the detection of the unmanned vehicle, and the radar module sensing system is determined according to the distribution location of the multiple radar modules, their corresponding detection range, and the current position of the unmanned vehicle. Multiple sensing events are identified based on the radar module sensing system, and a satellite radar control system is constructed based on each sensing event, the working data of each radar module, and the interaction events between each radar module. The dynamic control content of the unmanned vehicle is determined based on the satellite radar control system and the unmanned driving mode of the unmanned vehicle. The intelligent control events of the unmanned vehicle are determined based on the dynamic control content of the unmanned vehicle, the surrounding environment detected by the unmanned vehicle, and the working history of the satellite radar. In the intelligent control events of autonomous vehicles, multiple intelligent control items are identified based on the recognition of the intelligent control events of autonomous vehicles. The intelligent control status of autonomous vehicles is determined according to the content of each intelligent control item, the corresponding priority of the item, and the driving data of autonomous vehicles. The system collects data on the driving scenario in which the autonomous vehicle is located. Based on the autonomous vehicle's intelligent control status, the driving scenario, and past control events, it determines the current control command for the autonomous vehicle. Based on the current control command, environmental data detected by satellite radar, and obstacle data, it determines the corresponding intelligent driving content.

2. The intelligent control method for unmanned vehicles based on satellite radar according to claim 1, characterized in that, The method involves marking multiple radar modules based on the detection of the autonomous vehicle, and determining the radar module sensing system based on the distribution location of the multiple radar modules, their corresponding detection range, and the current position of the autonomous vehicle, including: Real-time monitoring and detection of unmanned vehicles; determination of multiple radar signals along different paths based on vehicle detection; identification of corresponding radar modules based on the tracing of each radar signal; marking of multiple radar modules. In multiple radar modules, the current position and corresponding detection range of each radar module are determined based on the identification of each radar module, and the distribution of multiple radar modules is determined based on the current position of each radar module; at the same time, the current position of the unmanned vehicle is collected, and the radar module sensing system is determined based on the distribution of multiple radar modules, the corresponding detection range, and the current position of the unmanned vehicle.

3. The intelligent control method for unmanned vehicles based on satellite radar according to claim 1, characterized in that, The process involves identifying multiple sensing events based on the radar module's sensing system, and constructing a satellite radar control system based on these sensing events, the operating data of each radar module, and the interaction events between the radar modules. This includes: The radar module sensing system is monitored in real time. Multiple sensing contents are determined based on the detection of the radar module sensing system. The sensing events of the radar module are determined according to each sensing content, the corresponding radar module and the working history of the radar module, so as to mark multiple sensing events. Multiple radar modules interact and the interaction events between each radar module are marked. At the same time, a satellite radar is constructed based on the distribution location, corresponding working data and corresponding functions of multiple radar modules. A control system for the satellite radar is constructed based on the radar architecture of the satellite radar, multiple sensing events and multiple radar modules.

4. The intelligent control method for unmanned vehicles based on satellite radar according to claim 1, characterized in that, The satellite radar-based control system and the autonomous driving mode of the unmanned vehicle determine the dynamic control content of the unmanned vehicle. Based on the dynamic control content of the unmanned vehicle, the surrounding environment detected by the unmanned vehicle, and the working history of the satellite radar, the intelligent control events of the unmanned vehicle are determined, including: Real-time monitoring of autonomous vehicle driving moments, collection of multiple autonomous driving data points during autonomous driving, and determination of autonomous driving mode based on multiple autonomous driving data points, autonomous vehicle driving behavior, and corresponding current location; Based on the identification of the satellite radar control system, multiple surrounding environmental features of the unmanned vehicle are determined. Based on these multiple surrounding environmental features, the unmanned driving mode of the unmanned vehicle, and the working status of the satellite radar itself, the dynamic control content of the unmanned vehicle is determined.

5. The intelligent control method for unmanned vehicles based on satellite radar according to claim 4, characterized in that, The satellite radar-based control system and the autonomous driving mode of the unmanned vehicle determine the dynamic control content of the unmanned vehicle. Based on the dynamic control content of the unmanned vehicle, the surrounding environment detected by the unmanned vehicle, and the working history of the satellite radar, the intelligent control events of the unmanned vehicle are determined. This also includes: The system collects data on the surrounding environment detected by the autonomous vehicle. Based on the surrounding environment detected by the autonomous vehicle and the dynamic control content of the autonomous vehicle, the first level of intelligent control content is determined. Based on the surrounding environment detected by the autonomous vehicle and the working history of the satellite radar, the second level of intelligent control content is determined. The intelligent control events of the autonomous vehicle are determined based on the first level of intelligent control content and the second level of intelligent control content.

6. The intelligent control method for unmanned vehicles based on satellite radar according to claim 1, characterized in that, In the intelligent control events of the autonomous vehicle, multiple intelligent control items are determined based on the identification of the intelligent control events. The intelligent control state of the autonomous vehicle is determined according to the content of each intelligent control item, its corresponding priority, and the vehicle's driving data, including: Real-time monitoring of intelligent control events of unmanned vehicles; identification of intelligent control events of unmanned vehicles to output multiple item tags; determination of corresponding intelligent control items based on the tracing of each item tag; collection of multiple intelligent control items.

7. The intelligent control method for unmanned vehicles based on satellite radar according to claim 6, characterized in that, In the intelligent control events of the autonomous vehicle, multiple intelligent control items are determined based on the identification of the intelligent control events of the autonomous vehicle. The intelligent control state of the autonomous vehicle is determined according to the item content, corresponding item priority, and driving data of the autonomous vehicle. This also includes: The content of each intelligent control project is determined based on the identification of each intelligent control project, and the priority of each intelligent control project is determined by comparison of each intelligent control project; at the same time, driving data of the unmanned vehicle is collected. The first intelligent control coefficient is determined based on the driving data of the autonomous vehicle and the content of each intelligent control project; the second intelligent control coefficient is determined based on the driving data of the autonomous vehicle and the priority of each intelligent control project; and the intelligent control state of the autonomous vehicle is determined according to the mapping relationship between the first intelligent control coefficient, the second intelligent control coefficient and the intelligent control state.

8. The intelligent control method for unmanned vehicles based on satellite radar according to claim 1, characterized in that, The process involves collecting data on the autonomous vehicle's driving scenario, determining the vehicle's current control commands based on its intelligent control status, the driving scenario, and past control events, and then determining the corresponding intelligent driving content based on the current control commands, environmental data detected by satellite radar, and obstacle data. This includes: The current position of the autonomous vehicle is marked, and multiple driving and environmental factors are determined based on the surrounding detection of the autonomous vehicle. The driving scenario of the autonomous vehicle is determined based on the current position of the autonomous vehicle, multiple driving and environmental factors. The system collects the intelligent control status of the autonomous vehicle, determines the first level of command content based on the intelligent control status and the driving scenario in which the autonomous vehicle is located, and collects the previous control events of the autonomous vehicle. Based on the intelligent control status and the previous control events of the autonomous vehicle, it determines the second level of command content, and determines the current control command of the autonomous vehicle based on the first level of command content and the second level of command content.

9. The intelligent control method for unmanned vehicles based on satellite radar according to claim 8, characterized in that, The process of collecting data on the driving scenario of the unmanned vehicle, determining the current control command of the unmanned vehicle based on its intelligent control status, the driving scenario, and past control events, and determining the corresponding intelligent driving content based on the current control command, environmental data detected by satellite radar, and obstacle data, also includes: The system collects environmental data detected by satellite radar and determines corresponding obstacle data based on the dynamic detection of satellite radar. It then determines the first level of driving content based on the current control command of the unmanned vehicle and the environmental data detected by satellite radar, determines the second level of driving content based on the current control command of the unmanned vehicle and the obstacle data detected by satellite radar, and finally determines the corresponding intelligent driving content based on the first and second levels of driving content.

10. An intelligent control system for an unmanned vehicle based on satellite radar, characterized in that, The intelligent control system for the satellite radar-based unmanned vehicle is applied to the intelligent control method for the satellite radar-based unmanned vehicle as described in any one of claims 1-9, wherein the intelligent control system for the satellite radar-based unmanned vehicle includes: The radar module sensing system module is used to mark multiple radar modules based on the detection of unmanned vehicles. The radar module sensing system is determined according to the distribution of multiple radar modules, their corresponding detection ranges, and the current position of the unmanned vehicle. The control system module is used to identify multiple sensing events based on the radar module's sensing system, and to construct a control system for the satellite radar based on each sensing event, the working data of each radar module, and the interaction events between each radar module. The intelligent control event module is used to determine the dynamic control content of the unmanned vehicle based on the satellite radar control system and the unmanned driving mode of the unmanned vehicle. The module determines the intelligent control events of the unmanned vehicle based on the dynamic control content of the unmanned vehicle, the surrounding environment detected by the unmanned vehicle, and the working history of the satellite radar. The intelligent control status module is used to identify multiple intelligent control items based on the intelligent control events of the autonomous vehicle, and to determine the intelligent control status of the autonomous vehicle based on the content of each intelligent control item, the corresponding priority of the item, and the driving data of the autonomous vehicle. The intelligent driving content module is used to collect the driving scene where the unmanned vehicle is located, determine the current control command of the unmanned vehicle based on the intelligent control status of the unmanned vehicle, the driving scene where the unmanned vehicle is located, and the previous control events of the unmanned vehicle, and determine the corresponding intelligent driving content based on the current control command of the unmanned vehicle, environmental data detected by satellite radar and obstacle data.

Images