Formation control method and device and intelligent driving equipment

Abstract

The invention discloses a formation control method and device and intelligent driving equipment, and belongs to the technical field of intelligent driving. According to the method, the formation control device can adjust the working frequency of the sensor on the movable object in the formation based on the current state data of the formation, so that the power consumption of the formation can be flexibly adjusted based on the current state data, and the problem that the power consumption of the formation is too large in some scenes is avoided. Besides, after the working frequency of the sensors in the formation is adjusted, for example, the working frequency of the sensors is reduced, the data collected by the sensors is reduced, and correspondingly, the data needing to be processed by the formation is also reduced, so that the calculation resources of the formation are saved.

Description

Formation control method, device and intelligent driving equipment Technical Field

[0001] The present application relates to the field of intelligent driving technology, and in particular to a formation control method, apparatus, and intelligent driving equipment. Background Art

[0002] In intelligent driving, platooning refers to the formation of multiple movable objects, such as vehicles, in a pre-planned formation. Each movable object is equipped with sensors such as image sensors (e.g., cameras and infrared sensors) and radar sensors (e.g., lidar, millimeter-wave radar, and ultrasonic radar). As each movable object in the platoon moves, the data collected by these sensors can be used to control its direction and speed, ensuring that the objects maintain their formation.

[0003] Summary of the Invention

[0004] The present application provides a formation control method, apparatus, and intelligent driving device, which can flexibly adjust the power consumption of a formation in intelligent driving.

[0005] In a first aspect, a formation control method is provided. In this method, current state data of a formation is obtained, where the formation includes M movable objects, where M is a positive integer greater than or equal to 2; a target sensor configuration is determined based on the current state data, where the target sensor configuration includes sensor operating frequency information for N of the M movable objects, and the target sensor configuration is capable of achieving safe movement of the formation, where N is less than or equal to M; and corresponding sensor operating frequency information is issued to each of the N movable objects, so that the movable objects adjust the operating frequency of their sensors based on the corresponding sensor operating frequency information.

[0006] In this application, the operating frequency of sensors on movable objects within the formation can be adjusted based on the formation's current state data. This allows for flexible adjustment of the formation's power consumption based on the current state data, thereby avoiding the problem of excessive power consumption in certain scenarios. Furthermore, by adjusting the operating frequency of sensors within the formation, such as lowering it, the data collected by the sensors will be reduced, and accordingly, the data that the formation needs to process will also be reduced, thereby conserving the formation's computing resources.

[0007] Optionally, determining the target sensor configuration based on the current state data may be implemented by determining the target sensor configuration from a plurality of candidate sensor configurations based on the current state data.

[0008] In this application, multiple candidate sensor configurations can be pre-set to facilitate rapid determination of the target sensor configuration based on current state data, thereby improving formation control efficiency.

[0009] Optionally, in this method, multiple initial sensor configurations can also be determined, each of the multiple initial sensor configurations including the operating frequency of the sensor of each movable object in the M movable objects; the safety level of each initial sensor configuration in the multiple initial sensor configurations is determined, the safety level indicating the possibility that the simulated formation can travel safely after being set up according to the initial sensor configuration; based on the safety level of each initial sensor configuration in the multiple initial sensor configurations, multiple candidate sensor configurations are determined from the multiple initial sensor configurations.

[0010] In this application, candidate sensor configurations can be screened based on the safety levels of different initial sensor configurations to ensure that subsequent formations can travel safely after operating according to a certain candidate sensor configuration.

[0011] Optionally, the implementation method for determining the safety level of each initial sensor configuration among multiple initial sensor configurations can be: for any initial sensor configuration among the multiple initial sensor configurations, setting the sensors of each simulated object in the simulated formation according to any initial sensor configuration, and making the simulated formation travel in the scenario use case of each simulated scene in multiple simulated scenes; based on the driving conditions of the simulated formation in the scenario use case of each simulated scene, determining the number of scenario use cases in which the simulated formation can travel safely in the multiple simulated scenes; and determining the ratio between the number of scenario use cases that can travel safely and the total number of scenario use cases in the multiple simulated scenes as the safety level of any initial sensor configuration.

[0012] In this application, candidate sensor configurations can be determined by simulating the driving conditions of the formation in a scenario use case of the simulated scenario. In this way, multiple candidate sensor configurations can be quickly trained.

[0013] Optionally, the method for determining the target sensor configuration from multiple candidate sensor configurations based on the current state data can be: inputting the current state data into a classification model so that the classification model outputs the target sensor configuration; the classification model is used to classify the input data into one of the multiple candidate sensor configurations and output the classification result.

[0014] In the present application, a classification model may also be pre-trained to quickly select a target sensor configuration that matches the current state data from a plurality of candidate sensor configurations through the classification model.

[0015] Optionally, the classification model includes a minimum regression decision tree; in the method, multiple sample state data can also be obtained, each sample state data indicates a state of the simulated formation; an initialized minimum regression decision tree is constructed based on the multiple sample state data, the initialized minimum regression decision tree includes multiple initial classification results, the number of the multiple initial classification results is the same as the number of the multiple candidate sensor configurations, and different initial classification results indicate different state complexities of the simulated formation; the multiple initial classification results are bound to the multiple candidate sensor configurations one by one to obtain a minimum regression decision tree; wherein, the greater the state complexity level indicated by an initial classification result among the multiple initial classification results, the higher the sensor operating frequency in the candidate sensor configuration bound to the corresponding initial classification result.

[0016] The classification model provided by this application can be implemented through the minimum regression decision tree, which improves the application flexibility of this application.

[0017] Optionally, the current state data includes X state weights corresponding one-to-one to the X states, each of the X state weights indicating the complexity of the corresponding state, where X is greater than or equal to 1. In this scenario, determining the target sensor configuration based on the current state data may be implemented by determining the target sensor configuration based on the X state weights.

[0018] For the states of different dimensions of the formation, the higher the complexity of the state in each dimension, the greater the impact on the power consumption of the formation. Therefore, the complexity of the state in each dimension can be quantified to obtain the state weight of the state in each dimension, so as to facilitate the subsequent adjustment of the sensor configuration of the formation through the quantified state weight.

[0019] Optionally, the X states include one or more of an environment state, a task state, and a load state;

[0020] Among them, the state weight corresponding to the environmental state includes the environmental state weight of at least one movable object in the formation, and the environmental state weight indicates the complexity of the environmental state around the corresponding movable object. The state weight corresponding to the task state indicates the complexity of the task currently performed by the leader in the formation. The state weight corresponding to the load state includes the load state weight of each movable object in the formation, and the load state weight indicates the complexity of the load of the corresponding movable object.

[0021] The more complex the environment surrounding the formation, the more tasks the formation needs to handle to cope with the complex environment, and accordingly, the greater the power consumption of the formation. Therefore, the complexity of the environment surrounding the movable objects in the formation has a greater impact on the overall power consumption of the formation. In addition, the leader of the formation, as the computing center of the entire formation, needs to handle a large number of tasks. Therefore, the complexity of the task currently performed by the leader in the formation has a greater impact on the overall power consumption of the formation. In addition, the load complexity of each movable object in the formation has a direct impact on the current power consumption of the movable object. Based on the above reasons, it is possible to consider adjusting the sensor configuration of the formation through environmental state weights, task state weights, and load state weights to reduce the overall power consumption of the formation.

[0022] Optionally, in this method, current status information of the sensor of each of the M movable objects can also be obtained; the overall energy efficiency ratio of the formation is determined based on the current status information of the sensor of each of the M movable objects; if the overall energy efficiency ratio is lower than the reference energy efficiency ratio, the roles of at least two movable objects in the formation are switched.

[0023] In the present application, the roles of movable objects in the formation may also be adjusted according to the energy efficiency ratio of the formation to further reduce the power consumption of the formation.

[0024] Optionally, if the overall energy efficiency ratio is lower than the reference energy efficiency ratio, the implementation method of switching the roles of at least two movable objects in the formation can be: if the overall energy efficiency ratio is lower than the reference energy efficiency ratio, determine the expected power consumption of the switching process, and the switching process is the process of switching the roles of at least two movable objects in the formation; determine the power difference between the power before switching and the power after switching of the formation; if the power difference exceeds the expected power consumption, switch the roles of at least two movable objects in the formation.

[0025] If the power difference exceeds the expected power consumption, it indicates that the switching process is relatively inexpensive, and in this case, the roles of the at least two movable objects in the formation can be switched. Correspondingly, if the power difference is lower than the expected power consumption, it indicates that the switching process is relatively expensive, and in this case, the roles of the at least two movable objects in the formation can be switched, thereby avoiding the need for further power consumption.

[0026] Optionally, the implementation method for determining the expected power consumption of the switching process can be: determining at least one task that needs to be performed during the switching process; determining the expected power consumption based on the empirical power of each task in the at least one task, where the empirical power indicates the power required when the corresponding task was executed before the current time.

[0027] By using the above method, the expected power consumption of the switching process can be determined, thereby improving the application flexibility of the present application.

[0028] In a second aspect, a formation control device is provided, which has the function of implementing the formation control method of the first aspect. The device includes at least one unit, which is configured to implement the formation control method provided in the first aspect and its implementations.

[0029] In a third aspect, another formation control device is provided, comprising a processor and a memory, wherein the memory is configured to store a program that supports the formation control device in executing the formation control method provided in the first aspect and its implementations, and to store data related to implementing the formation control method provided in the first aspect and its implementations. The processor is configured to execute the program stored in the memory. The formation control device may further include a communication bus for communication between the processor and the memory.

[0030] In a fourth aspect, an intelligent driving device is provided, the intelligent driving device including the formation control apparatus described in the second or third aspect. The intelligent driving device may be a vehicle, including a passenger car, a commercial vehicle, or an engineering vehicle.

[0031] In a fifth aspect, a computer-readable storage medium is provided, wherein instructions are stored in the computer-readable storage medium. When the computer-readable storage medium is run on a computer, the computer executes the formation control method described in the first aspect and its various implementation methods.

[0032] In a sixth aspect, a computer program product comprising instructions is provided, which, when executed on a computer, enables the computer to execute the formation control method described in the first aspect and its various implementations.

[0033] The technical effects obtained in the above-mentioned second to sixth aspects are similar to the technical effects obtained by the corresponding technical means in the first aspect, and will not be repeated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] FIG1 is a schematic structural diagram of an intelligent driving device 100 provided in an embodiment of the present application;

[0035] FIG2 is an internal structure diagram of the computer system 112 shown in FIG1 provided in an embodiment of the present application;

[0036] FIG3 is a schematic diagram of a formation provided in an embodiment of the present application;

[0037] FIG4 is a schematic diagram of the architecture of a formation control system provided in an embodiment of the present application;

[0038] FIG5 is a schematic diagram of the architecture of another formation control system provided in an embodiment of the present application;

[0039] FIG6 is a flow chart of a formation control method provided in an embodiment of the present application;

[0040] FIG7 is a schematic diagram of a simulation scenario provided in an embodiment of the present application;

[0041] FIG8 is a schematic diagram of a process for obtaining multiple candidate sensor configurations through training according to an embodiment of the present application;

[0042] FIG9 is a schematic diagram of the structure of a minimum regression decision tree provided in an embodiment of the present application;

[0043] FIG10 is a schematic diagram of a process of switching formation roles provided in an embodiment of the present application;

[0044] FIG11 is a schematic diagram of a scenario of switching formation roles provided in an embodiment of the present application;

[0045] FIG12 is a schematic diagram of another process of switching formation roles provided in an embodiment of the present application;

[0046] FIG13 is a schematic structural diagram of a formation control device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0047] In order to make the objectives, technical solutions and advantages of this application clearer, the embodiments of this application will be further described in detail below with reference to the accompanying drawings.

[0048] Before explaining the embodiments of the present application in detail, the application scenarios of the embodiments of the present application are first explained.

[0049] Intelligent driving refers to the use of sensors, controllers, actuators, communication modules, and other devices configured on intelligent driving devices to assist the driver in controlling the intelligent driving device, or to replace the driver in achieving unmanned driving. Based on this, intelligent driving can include assisted driving, autonomous driving, and unmanned driving. Examples of intelligent driving devices include vehicles or drones.

[0050] When multiple intelligent driving devices have similar destinations and travel routes, they can form a platoon and travel in formation. For example, if multiple trucks have the same departure and destination, they can be formed into a platoon and then controlled to travel in formation.

[0051] A formation typically consists of two roles: a leader and followers. The leader, acting as the core computing unit, receives sensor data from each follower via network communications, acquires data from its own sensors, processes this data, and then transmits information such as speed and direction to the followers to complete formation control.

[0052] The operating frequencies of all sensors on each intelligent driving device in the formation can be preconfigured. During platooning, all sensors on each intelligent driving device operate at the preconfigured operating frequencies. However, this approach consumes a lot of power and wastes computing resources. Based on this, an embodiment of the present application provides a formation control method that can flexibly adjust the operating frequencies of the sensors on each intelligent driving device in the formation. This method can avoid excessive power consumption in certain scenarios while also saving computing resources.

[0053] The hardware architecture of the intelligent driving device involved in the embodiments of the present application is explained below.

[0054] FIG1 is a schematic diagram of the structure of an intelligent driving device 100 provided in an embodiment of the present application. As shown in FIG1 , the intelligent driving device 100 may include multiple subsystems, such as the travel system 102, the sensing system 104, the control system 106, one or more peripheral devices 108, as well as the power supply 110, the computer system 112, and the user interface 116 in FIG1 . Optionally, the intelligent driving device 100 may include more or fewer subsystems, and each subsystem may include multiple components, which are not illustrated one by one here. In addition, each subsystem and component of the intelligent driving device 100 may be interconnected via wired or wireless connections.

[0055] The functions of each subsystem in the intelligent driving device 100 are described below.

[0056] (1) Travel system 102

[0057] The travel system 102 may include components that provide powered motion for the intelligent driving device 100. In some embodiments, the travel system 102 may include an engine 118, an energy source 119, a transmission 120, and wheels / tires 121.

[0058] The engine 118 can be an internal combustion engine, an electric motor, an air compression engine, or a combination of other types of engines, such as a hybrid engine consisting of a gasoline engine and an electric motor, or a hybrid engine consisting of an internal combustion engine and an air compression engine. The engine 118 converts the energy source 119 into mechanical energy.

[0059] Examples of energy source 119 include gasoline, diesel, other petroleum-based fuels, propane, other compressed gas-based fuels, ethanol, solar panels, batteries, and other sources of electricity. Energy source 119 can also provide energy for other systems of the intelligent driving device 100.

[0060] The transmission 120 can transmit mechanical power from the engine 118 to the wheels 121. The transmission 120 may include a gearbox, a differential, and a drive shaft. In some embodiments, the transmission 120 may also include other components, such as a clutch. The drive shaft may include one or more shafts that can be coupled to one or more wheels 121.

[0061] (2) Sensing system 104

[0062] The sensing system 104 may include several sensors that sense information about the environment surrounding the intelligent driving device 100. For example, the sensing system 104 may include a positioning system 122 (the positioning system may be a global positioning system (GPS), a BeiDou system, or other positioning systems), an inertial measurement unit (IMU) 124, a radar sensor 126 (the radar sensor may be a lidar, a millimeter-wave radar, or an ultrasonic radar, etc.), a laser rangefinder 128, and an image sensor 130 (the image sensor may be a camera or an infrared sensor). The sensing system 104 may also include sensors of the internal systems of the monitored intelligent driving device 100 (for example, an in-vehicle air quality monitor, a fuel gauge, an oil temperature gauge, etc.). Sensor data from one or more of these sensors can be used to detect objects and their corresponding characteristics (position, shape, direction, speed, etc.). This detection and recognition is a key function for the safe operation of the intelligent driving device 100.

[0063] The positioning system 122 may be used to estimate the geographic location of the intelligent driving device 100. The IMU 124 is used to sense the position and orientation changes of the intelligent driving device 100 based on inertial acceleration. In some embodiments, the IMU 124 may be a combination of an accelerometer and a gyroscope.

[0064] The radar sensor 126 may utilize radio signals to sense objects in the surrounding environment of the intelligent driving device 100. In some embodiments, in addition to sensing objects, the radar sensor 126 may also be used to sense the speed and / or heading of the objects.

[0065] The laser rangefinder 128 may utilize laser light to sense objects in the environment in which the intelligent driving device 100 is located. In some embodiments, the laser rangefinder 128 may include one or more laser sources, a laser scanner, and one or more detectors, among other system components.

[0066] The image sensor 130 may be used to capture multiple images of the surrounding environment of the intelligent driving device 100. The image sensor 130 may be a still camera or a video camera. The number of the image sensor 130 may be one or more.

[0067] (3) Control system 106

[0068] The control system 106 controls the operation of the intelligent driving device 100 and its components and may include various components, including, for example, a steering system 132 , a throttle 134 , a braking unit 136 , a fusion algorithm unit 138 , a computer vision system 140 , a path control system 142 , and an obstacle avoidance system 144 .

[0069] The steering system 132 can adjust the forward direction of the intelligent driving device 100. In some embodiments, the steering system 132 can be a steering wheel system.

[0070] The throttle 134 is used to control the operating speed of the engine 118 and thus the speed of the intelligent driving device 100 .

[0071] Braking unit 136 is used to control the deceleration of intelligent driving device 100. Braking unit 136 can use friction to slow down wheel 121. In other embodiments, braking unit 136 can convert the kinetic energy of wheel 121 into electric current. Braking unit 136 can also take other forms to slow the rotation speed of wheel 121 and thus control the speed of intelligent driving device 100.

[0072] The fusion algorithm unit 138 is used to collect and comprehensively analyze the data obtained by multiple sensors in the sensing system 104 so as to more accurately and reliably describe the external environment, thereby improving the driving safety of the intelligent driving device.

[0073] The computer vision system 140 can process and analyze images captured by the image sensor 130 to identify objects and / or features in the environment surrounding the intelligent driving device 100, and / or process and analyze data information captured by the radar sensor 126. The objects and / or features may include traffic signs, road boundaries, and obstacles. The computer vision system 140 can use object recognition algorithms, structure from motion (SFM) algorithms, video tracking, and other computer vision technologies. In some embodiments, the computer vision system 140 can be used to map the environment, track objects, estimate the speed of objects, and so on.

[0074] The route control system 142 is used to determine the driving route of the intelligent driving device 100. In some embodiments, the route control system 142 can combine data from the sensor system 104 and one or more predetermined maps to determine the driving route for the intelligent driving device 100.

[0075] The obstacle avoidance system 144 is used to identify, assess, and avoid or otherwise negotiate potential obstacles in the environment of the intelligent driving device 100 .

[0076] Of course, in some examples, the control system 106 may include additional or alternative components other than those shown and described, or may reduce some of the components shown above, which will not be illustrated here one by one.

[0077] (4) Peripheral equipment

[0078] The intelligent driving device 100 interacts with external sensors, other vehicles, other computer systems, or users through peripheral devices 108. The peripheral devices 108 may include a wireless communication system 146, an onboard computer 148, a microphone 150, and / or a speaker 152.

[0079] In some embodiments, the peripheral device 108 provides a means for the user of the intelligent driving device 100 to interact with the user interface 116. For example, the onboard computer 148 can provide information to the user of the intelligent driving device 100. The user interface 116 can also operate the onboard computer 148 to receive user input. The onboard computer 148 can be operated via a touch screen. In other cases, the peripheral device 108 can provide a means for the intelligent driving device 100 to communicate with other devices located in the vehicle. For example, the microphone 150 can receive audio (e.g., voice commands or other audio input) from the user of the intelligent driving device 100. Similarly, the speaker 152 can output audio to the user of the intelligent driving device 100.

[0080] The wireless communication system 146 can communicate wirelessly with one or more devices directly or via a communication network. For example, the wireless communication system 146 can use third-generation (3G) cellular communications, such as code division multiple access (CDMA), global system for mobile communications (GSM), or general packet radio service (GPRS), or 4G cellular communications, such as long term evolution (LTE), or 5G cellular communications. The wireless communication system 146 can communicate with a wireless local area network (WLAN) using wireless fidelity (Wi-Fi). In some embodiments, the wireless communication system 146 can communicate directly with the device using an infrared link, Bluetooth, or ZigBee. The wireless communication system 146 may include one or more dedicated short range communication (DSRC) devices, which may include public and / or private data communications between vehicles and / or roadside stations.

[0081] (5) Power supply 110

[0082] The power supply 110 can provide power to the various components of the intelligent driving device 100. In some embodiments, the power supply 110 can be a rechargeable lithium-ion or lead-acid battery. One or more battery packs of such batteries can be configured as a power source to provide power to the various components of the intelligent driving device 100. In some embodiments, the power supply 110 and the energy source 119 can be implemented together.

[0083] (6) Computer System 112

[0084] Some or all functions of the intelligent driving device 100 are controlled by a computer system 112. The computer system 112 may include at least one processor 113 that executes instructions 115 stored in a non-transitory computer-readable medium, such as a memory 114. The computer system 112 may also be a plurality of computing devices that control individual components or subsystems of the intelligent driving device 100 in a distributed manner.

[0085] The processor 113 may be any conventional processor, such as a central processing unit (CPU). Alternatively, the processor may be an application-specific integrated circuit (ASIC) or other hardware-based processor.

[0086] In some embodiments, the memory 114 may include instructions 115 (e.g., program logic) that may be executed by the processor 113 to perform various functions of the intelligent driving device 100, including those described above. The memory 114 may also include additional instructions, including instructions for sending data to, receiving data from, interacting with, and / or sending control instructions to one or more of the travel system 102, the sensing system 104, the control system 106, and the peripheral devices 108.

[0087] In addition to instructions 115, memory 114 may also store data, such as point cloud data, image data, and other information. This information may be used by computer system 112 during operation of intelligent driving device 100 in autonomous, semi-autonomous, and / or manual modes.

[0088] (7) User Interface 116

[0089] The user interface 116 is used to provide information to or receive information from a user of the intelligent driving device 100. Optionally, the user interface 116 may include one or more input / output devices within the set of peripheral devices 108, such as a wireless communication system 146, an onboard computer 148, a microphone 150, and a speaker 152.

[0090] Based on the functions of the various subsystems described above, the computer system 112 can control the functions of the intelligent driving device 100 based on input information received from the various subsystems and from the user interface 116. For example, the computer system 112 can use input information from the control system 106 to control the steering system 132 to avoid obstacles detected by the sensing system 104 and the obstacle avoidance system 144.

[0091] Optionally, one or more of the above components may be installed or associated separately from the intelligent driving device 100. For example, the memory 114 may be partially or completely separate from the intelligent driving device 100. The above components may be communicatively coupled together in a wired and / or wireless manner.

[0092] Optionally, the above components are only an example. In actual applications, the components in the above modules may be added or deleted according to actual needs. Figure 1 should not be understood as a limitation to the embodiments of the present application.

[0093] The intelligent driving device 100 may include road vehicles, water vehicles, air vehicles, industrial equipment, agricultural equipment, or entertainment equipment. For example, the intelligent driving device may be a vehicle, which is a vehicle in a broad sense, and may be a transportation vehicle (such as a commercial vehicle, a passenger car, a motorcycle, a flying car, a train, etc.), an industrial vehicle (such as a forklift, a trailer, a tractor, etc.), an engineering vehicle (such as an excavator, a bulldozer, a crane, etc.), agricultural equipment (such as a lawn mower, a harvester, etc.), amusement equipment, a toy vehicle, etc. This application does not specifically limit the type of vehicle. For another example, the intelligent driving device may be a vehicle such as an airplane or a ship.

[0094] It is understandable that the structure of the intelligent driving device in FIG1 is only an exemplary embodiment, and the intelligent driving device in the embodiments of the present application includes but is not limited to the above structure.

[0095] FIG2 is an internal structure diagram of the computer system 112 shown in FIG1 provided in an embodiment of the present application.

[0096] 2 , the computer system 112 may include a processor 113 coupled to a system bus 201. The processor 113 may be one or more processors, each of which may include one or more processor cores.

[0097] Processor 113 may be any conventional processor, including a reduced instruction set computer (RISC), a complex instruction set computer (CISC), or a combination thereof. Alternatively, the processor may be an application-specific integrated circuit (ASIC). Alternatively, processor 113 may be a neural network processor, or a combination of a neural network processor and the conventional processors described above.

[0098] Alternatively, in various embodiments described herein, the computer system 112 may be located remotely from the intelligent driving device and may communicate wirelessly with the intelligent driving device. In other aspects, some of the processes described herein are executed on a processor within the intelligent driving device, while others are executed by a remote processor.

[0099] As shown in Figure 2, the computer system 112 may further include a memory 114 coupled to the system bus 201. The memory 114 is used to store information, such as the instruction 115 in Figure 1, which may be used to implement the formation control method provided in the embodiment of the present application.

[0100] As shown in FIG. 2 , the computer system 112 may further include a display adapter 202 . The display adapter 202 may drive a display 203 . The display adapter 202 is coupled to a system bus 201 .

[0101] As shown in FIG2 , the system bus 201 may also be coupled to an input / output (I / O) bus 205 via a bus bridge 204. An I / O interface 206 is coupled to the I / O bus 205. The I / O interface 206 communicates with various I / O devices. Examples of the various I / O devices include an input device 207 (e.g., a keyboard, a mouse, a touch screen, etc.), a media tray 208 (e.g., a compact disc read-only memory (CD-ROM), a multimedia interface, etc.), a transceiver 209 (capable of sending and / or receiving radio communication signals), a sensor 210 (which may be any of the sensors in the sensor system 104 in FIG1 ), and an external universal serial bus (USB) interface 211.

[0102] In addition, as shown in FIG2 , computer system 112 may further include a network interface 212, through which computer system 112 communicates with software deployment server 213. Network interface 212 is a hardware network interface, such as a network card. Network 214 between network interface 212 and software deployment server 213 may be an external network, such as the Internet, or an internal network, such as Ethernet or a virtual private network (VPN). Alternatively, network 214 may be a wireless network, such as a wireless fidelity (Wi-Fi) network, a cellular network, or the like.

[0103] 2 , the computer system 112 may further include a hard disk drive interface 215 coupled to the system bus 201 . The hard disk drive interface 215 is connected to a hard disk drive 216 .

[0104] 2 , the computer system 112 may further include a system memory 217 coupled to the system bus 201 . Data running in the system memory 217 may include an operating system (OS) 218 ​​and application programs 219 of the computer system 112 .

[0105] The operating system 217 includes a shell 220 and a kernel 221. Shell 220 serves as an interface between the user and the operating system's kernel. Shell 220 manages the interaction between the user and the operating system, such as waiting for user input, interpreting user input to the operating system, and processing various operating system outputs.

[0106] The kernel 221 consists of the operating system's components responsible for managing memory, files, peripherals, and system resources. The kernel 221 directly interacts with the hardware. It typically runs processes and provides inter-process communication, central processing unit (CPU) time slice management, interrupt management, memory management, and I / O management.

[0107] The application 219 includes programs 222 related to controlling the intelligent driving of the intelligent driving device, such as programs for managing the interaction between the intelligent driving car and obstacles on the road, programs for controlling the route or speed of the automatic intelligent car, and programs for controlling the interaction between the intelligent driving car and other intelligent driving cars on the road.

[0108] The application 219 may also exist on a system of a software deploying server 213 . For example, when the intelligent driving related program 222 needs to be executed, the computer system 112 may download the intelligent driving related program 222 from the software deploying server 213 .

[0109] It is understandable that the computer system 112 of the intelligent driving device in FIG2 is only an exemplary implementation. The computer system 112 applied to the intelligent driving device in the embodiment of the present application includes but is not limited to the above structure.

[0110] Multiple intelligent driving devices based on those shown in FIG. 1 and FIG. 2 can form a formation and then travel in the form of a formation.

[0111] Figure 3 is a schematic diagram of a formation provided in an embodiment of the present application. As shown in Figure 3, the formation includes multiple movable objects, labeled A, B, and C in Figure 3. Each of these multiple movable objects can be, for example, the intelligent driving device shown in Figure 1. Figure 3 uses three intelligent driving vehicles as an example for illustration.

[0112] Furthermore, any two of these multiple movable objects can communicate and exchange data, enabling the multiple movable objects to travel in a platoon. As shown in Figure 3, among the three intelligent driving vehicles, intelligent driving vehicle A serves as the leader (or lead vehicle) of the platoon, and the other two intelligent driving vehicles serve as followers (or following vehicles).

[0113] To implement the formation control method provided in the embodiments of this application, the embodiments of this application also provide a formation control system. Figure 4 is a schematic diagram of the architecture of a formation control system provided in the embodiments of this application. As shown in Figure 4, the formation control system includes a traffic complexity assessment module 401, a task density assessment module 402, a load assessment module 403, a formation power consumption adjustment module 404, and a dynamic sensor configuration module 405.

[0114] Among them, the traffic complexity assessment module 401 is used to assess the complexity of the state of the environment surrounding the movable object, and based on the assessed complexity of the surrounding environment, determine the environmental state weight corresponding to the environmental state. In some embodiments, as shown in Figure 4, the traffic complexity assessment module 401 can assess the complexity of the state of the environment surrounding the movable object based on the surround detection results of the movable object and real-time high-definition map information. The specific implementation method is described in detail in the subsequent embodiments. Among them, the surround detection results of the movable object can include, for example, the density of obstacles in the environment surrounding the movable object. The surround detection results can be determined based on image data and / or point cloud data collected by the movable object.

[0115] The more complex the environment around the formation, the more tasks the formation needs to handle to cope with the complex environment, and accordingly, the greater the power consumption of the formation. Therefore, the complexity of the environment surrounding the movable objects in the formation has a greater impact on the overall power consumption of the formation. Based on this, in order to be able to flexibly adjust the power consumption of the formation, the traffic complexity assessment module 401 can be deployed on each movable object to assess the complexity of the environment surrounding each movable object. Optionally, considering that the environment surrounding the followers in the formation has a greater impact on the formation, for the formation shown in Figure 3, the traffic complexity assessment module 401 can be deployed on each follower in the formation, and the leader of the formation is not deployed with the traffic complexity assessment module 401.

[0116] Task density assessment module 402 is used to assess the complexity of the task currently being performed by the team leader in the formation, and based on the complexity of the task being assessed, determines the task state weight corresponding to the task state. In certain embodiments, task density assessment module 402 can assess the complexity of the task based on the number of tasks (or task queue length) in the task queue currently being performed, and specific implementation methods are described in detail in subsequent embodiments. As shown in Figure 4, the tasks in the task queue include, for example, formation path planning tasks, follow-up control adjustment tasks, and anti-collision / lane keeping tasks, etc.

[0117] The leader of a formation serves as the computing center for the entire formation and needs to process a large number of tasks. Therefore, the complexity of the task currently being performed by the leader has a significant impact on the overall power consumption of the formation. Therefore, to flexibly adjust the power consumption of the formation, a task density assessment module 402 is deployed on the leader of the formation to assess the complexity of the task currently being performed by the leader.

[0118] The load evaluation module 403 is configured to evaluate the complexity of the load on the movable object and, based on the evaluated load complexity, determine a load state weight corresponding to the load state. In some embodiments, as shown in FIG4 , the load state can be determined based on sensor status on the movable object, such as sensor power consumption and temperature, and data such as the usage rate, memory usage, and communication bandwidth usage of the graphics processing unit (GPU) or CPU on the movable object. The specific implementation method will be described in detail in subsequent embodiments.

[0119] The load complexity of each movable object in the formation has a direct impact on the current power consumption of the movable object. Based on this, for the formation shown in Figure 3, the load evaluation module 403 can be deployed on each movable object to evaluate the load complexity of each movable object.

[0120] The formation power consumption adjustment module 404 is configured to determine a target sensor configuration for the formation based on the environment state weight determined by the traffic complexity assessment module 401, the task state weight determined by the task density assessment module 402, and the load state weight determined by the load assessment module 403. The target sensor configuration includes the operating frequency of the sensor on each movable object in at least one movable object in the formation.

[0121] Wherein, the target sensor configuration may exemplarily include the sensor operating frequency information of each movable object among all movable objects in the formation. Optionally, it may also include the sensor operating frequency information of each movable object in some movable objects in the formation. Wherein, the sensor operating frequency information of any movable object may exemplarily include the operating frequency of each sensor deployed on the movable object, and optionally may also include the operating frequency of some sensors deployed on the movable object. In addition, the sensor that adjusts the operating frequency in the embodiment of the present application refers to a configurable sensor deployed on the movable object (that is, the operating frequency can be adjusted). For non-configurable sensors deployed on the movable object (that is, the operating frequency is not allowed to be adjusted), they are not within the scope of consideration of the formation control method provided in the embodiment of the present application.

[0122] In scenarios where the movable object is an intelligent driving vehicle, configurable sensors may include, for example, a front medium-range camera, a front short-range camera, a front long-range camera, a rearview camera, a top surround-view camera, a front millimeter-wave radar, a rear millimeter-wave radar, and a side millimeter-wave radar. Optionally, the range of configurable sensors can be modified based on the type of intelligent driving vehicle.

[0123] The formation power consumption adjustment module 404 can be deployed on any movable object in the formation. Alternatively, since the leader in the formation consumes a large amount of power, in order to avoid excessive power consumption of the leader, the formation power consumption adjustment module 404 can be deployed on any follower in the formation.

[0124] Dynamic sensor configuration module 405 is configured to adjust the operating frequency of a sensor on at least one movable object based on the target sensor configuration determined by formation power consumption adjustment module 404. This implements the technical solution provided by the embodiments of the present application for dynamically adjusting the operating frequency of sensors on movable objects in a formation, thereby enabling flexible adjustment of the formation's power consumption.

[0125] For example, as shown in Figure 4, the dynamic sensor configuration module 405 can turn off a sensor on a movable object (that is, the operating frequency is set to 0), or turn on a sensor on a movable object (that is, the operating frequency is set to the initial operating frequency), or adjust the operating frequency of a sensor on a movable object according to the target sensor configuration.

[0126] For example, the dynamic sensor configuration module 405 can be configured on each movable object in the formation to adjust the operating frequency of the sensor of any movable object in the formation. Optionally, during platooning, the leader's sensor is typically fully active and operating at maximum frequency to ensure safe platooning. Based on this, the dynamic sensor configuration module 405 can be configured on each follower in the formation to adjust the operating frequency of the sensor of any follower in the formation.

[0127] In addition, as shown in FIG4 , the formation control system may further include a classification model generation module 406. The classification model generation module 406 is used to pre-train a classification model so that the formation power consumption adjustment module 404 can output the target sensor configuration based on the three weights obtained through the trained classification model. The specific implementation method is described in detail in subsequent embodiments.

[0128] In some embodiments, the classification model generation module 406 includes a pre-trained configuration model, a scenario simulator, and a security verification module, etc., so as to pre-train the classification model through these modules. The detailed implementation method of the classification model generation module 406 will be described in detail in subsequent embodiments.

[0129] The classification model generation module 406 can be deployed on any movable object of the formation. Alternatively, the classification model generation module 406 can be deployed on a remote server to reduce unnecessary data calculations of the formation, thereby reducing the power consumption of the formation.

[0130] 4 , the formation control system may further include a dynamic formation adjustment module 407. The dynamic formation adjustment module 407 is used to adjust the roles of the movable objects in the formation, such as adjusting the original leader to a follower, or adjusting one of the original followers to a leader, to complete the formation role switching.

[0131] In some embodiments, the dynamic formation adjustment module 407 can determine whether to switch the formation role according to the energy efficiency ratio of the formation to achieve energy management of the formation. The specific implementation method will be described in detail in the subsequent embodiments.

[0132] The dynamic formation adjustment module 407 can be integrated on the same movable object as the formation power consumption adjustment module 404. Alternatively, the dynamic formation adjustment module 407 can be deployed on any movable object, which is not limited in the embodiment of the present application.

[0133] 4 , the formation control system may further include a task scheduling module 408. The task scheduling module is used to determine tasks to be performed by each movable object in the formation and to send the tasks to the corresponding movable objects.

[0134] The task scheduling module 408 is illustratively deployed on the leader of the formation, so that the leader determines the tasks to be performed by each follower in the formation and controls the followers to perform the corresponding tasks. The leader also determines the leader tasks to be performed by itself and controls itself to perform the leader tasks. In some embodiments, the tasks to be performed by each follower in the formation are generally the same, so as shown in Figure 4, such tasks can also be referred to as common tasks. Examples of common tasks include power consumption summary statistics for the entire formation, communication status monitoring for the entire formation, etc.

[0135] The various modules in the formation control system shown in FIG4 are for illustrative purposes only. The formation control module provided in the embodiment of the present application may optionally include more or fewer modules, which will not be illustrated one by one here.

[0136] FIG5 illustrates another platoon control system provided by an embodiment of the present application. Compared to the platoon control system shown in FIG4 , the platoon control system shown in FIG5 may not include the traffic complexity assessment module 401 and the dynamic platoon adjustment module 407. The functions of the various modules shown in FIG5 can be referenced to those of the corresponding modules in FIG4 and will not be repeated here.

[0137] The formation control system shown in FIG5 can be applied in scenarios where the surrounding environment of the formation is simple and stable, such as in unmanned mining areas. In these scenarios, the surrounding environment of the formation does not change significantly and has little impact on the overall power consumption of the formation, so it is not necessary to consider the surrounding environment of the formation. Therefore, in these scenarios, the traffic complexity assessment module 401 can be omitted from the formation control system.

[0138] In addition, in the scenario of an unmanned mining area, since the lanes where the formation travels are relatively narrow, there is basically no need to switch the roles of the movable objects in the formation. Therefore, in these scenarios, the dynamic formation adjustment module 407 can be omitted from the formation control system.

[0139] The formation control systems shown in Figures 4 and 5 are for illustration purposes only and do not limit the formation control method provided in the embodiments of the present application. The formation control method provided in the embodiments of the present application can also be applied to other types of formation control systems, which will not be further illustrated here.

[0140] In addition, the modules shown in Figures 4 and 5 are software modules. As described above, these modules can be deployed on various movable objects in the fleet. The deployment locations of the modules are provided for illustrative purposes only. In the embodiments of the present application, different modules can be deployed on different movable objects as needed.

[0141] Optionally, these software modules may also be uniformly deployed on a remote server, and the remote server may uniformly control the power consumption of the fleet, which is not limited in the embodiments of the present application.

[0142] FIG6 is a flow chart of a formation control method provided by an embodiment of the present application. As shown in FIG6 , the method includes the following steps 601 to 603 .

[0143] Step 601: Acquire current state data of a formation, where the formation includes M movable objects, where M is a positive integer greater than or equal to 2.

[0144] The M movable objects may be, for example, the intelligent driving devices shown in FIG1 , so as to realize formation control in intelligent driving through the method provided in the embodiment of the present application.

[0145] The current state data may be understood as state data that affects the current power consumption of the formation.

[0146] For example, for a formation of vehicles traveling on a road, the surrounding environment state of the formation has a greater impact on the power consumption of the formation, so the current state data may include data corresponding to the surrounding environment state of the formation.

[0147] As another example, the number of tasks in the task queue executed by the team leader in the formation has a greater impact on the power consumption of the formation, so the current state data may include data corresponding to the state of the current task execution of the team leader in the formation.

[0148] As another example, the load of each movable object in the formation has a significant impact on the power consumption of the formation, so the current state data may include data corresponding to the load state of each movable object in the formation.

[0149] In order to facilitate subsequent adjustment of the formation's sensor configuration based on the current state data, the formation's current state may be quantified to obtain quantized current state data, so that the formation's sensor configuration can be adjusted subsequently through the quantized current state data.

[0150] Based on this, in some embodiments, the current state data includes X state weights corresponding to the X states one by one, each of the X state weights indicates the complexity of the corresponding state, and X is greater than or equal to 1.

[0151] For the different dimensions of the formation's state, the more complex the state in each dimension, the greater the impact on the formation's power consumption. Therefore, the complexity of the state in each dimension can be quantified to obtain the state weight of each dimension. The larger the state weight, the more complex the corresponding state.

[0152] For example, the X states include one or more of an environment state, a task state, and a load state. The state weight corresponding to the environment state includes an environment state weight of at least one movable object in the formation, indicating the complexity of the environment surrounding the corresponding movable object; the state weight corresponding to the task state indicates the complexity of the task currently being performed by the leader in the formation; and the state weight corresponding to the load state includes a load state weight of each movable object in the formation, indicating the complexity of the load of the corresponding movable object.

[0153] The complexity of the state of the environment surrounding the movable object can be characterized by the density of obstacles in the environment surrounding the movable object. The greater the density of obstacles in the environment surrounding the movable object, the more complex the state of the environment surrounding the movable object. Based on this, in some implementations, the weight of the environment state of the movable object can be determined based on the density of obstacles in the environment surrounding the movable object.

[0154] For example, multiple obstacle density intervals and corresponding environment state weights are pre-set. When the obstacle density in the current surrounding environment of the movable object is determined, the environment state weight of the movable object can be determined based on the determined obstacle density and the corresponding relationship.

[0155] The density of obstacles in the current surroundings of the movable object can be determined using surround view detection results and real-time high-definition map information. In addition, obstacles in the surrounding environment include pedestrians and other intelligent driving devices other than the platoon.

[0156] In addition, the complexity of the task currently being performed by the team leader can be characterized by the length of the task queue currently being performed by the team leader. The longer the task queue currently being performed by the team leader is, the higher the complexity of the task currently being performed by the team leader is. Based on this, in some implementations, the task state weight of the team leader can be determined based on the length of the task queue currently being performed by the team leader.

[0157] The task status weight of the team leader is determined based on the length of the task queue currently executed by the team leader, which can also be determined through a pre-set corresponding relationship and will not be described in detail here.

[0158] The length of the task queue currently being executed by the team leader may be the number of tasks included in the task queue.

[0159] In addition, the complexity of the load of the movable object can be characterized by the power consumption of the sensors on the movable object, the CPU or GPU usage rate, the memory occupancy rate, and the communication bandwidth occupancy rate. The greater the power consumption of the sensors on the movable object, the greater the CPU or GPU usage rate, the greater the memory occupancy rate, and the greater the communication bandwidth occupancy rate, the higher the complexity of the load of the movable object. Based on this, in some implementations, the task state weight of the leader can be determined based on the current power consumption of the sensors on the movable object, the current usage rate of the CPU or GPU, the current occupancy rate of the memory, and the current occupancy rate of the communication bandwidth.

[0160] Among them, the task state weight of the leader is determined based on the current power consumption of the sensor on the movable object, the current usage rate of the CPU or GPU, the current occupancy rate of the memory, and the current occupancy rate of the communication bandwidth. It can also be determined through a pre-set corresponding relationship, which will not be repeated here.

[0161] Step 602: Determine a target sensor configuration based on current state data, where the target sensor configuration includes sensor operating frequency information of N movable objects among M movable objects, and the target sensor configuration can achieve safe driving of the platoon, where N is less than or equal to M.

[0162] In some embodiments, multiple candidate sensor configurations may be pre-set to improve formation control efficiency. In this scenario, step 602 may be implemented by determining a target sensor configuration from the multiple candidate sensor configurations based on current state data.

[0163] Determining a target sensor configuration from among multiple candidate sensor configurations based on current state data can be accomplished using a classification model. Specifically, the current state data is input into the classification model, causing the classification model to output the target sensor configuration. The classification model is configured to classify the input data into one of the multiple candidate sensor configurations and output a classification result.

[0164] The classification model may be any pre-trained model for classification and outputting classification results, and may include, for example, a minimum regression decision tree or other neural network models.

[0165] Alternatively, in embodiments of the present application, a target sensor configuration may be determined from multiple candidate sensor configurations without using a classification model. In this scenario, state data conditions for each of the multiple candidate sensor configurations may be preconfigured. If the current state data matches the state data conditions for a particular candidate sensor configuration, that candidate sensor configuration is selected as the target sensor configuration.

[0166] In a scenario where the current state data includes X state weights corresponding one-to-one to X states, the state data condition example may be a plurality of state weight ranges.

[0167] Optionally, in other embodiments, multiple candidate sensor configurations may not be pre-set, and when the current state data of the formation is acquired, the target sensor configuration may be calculated using a relevant algorithm.

[0168] In addition, in some embodiments, in a scenario where the current state data includes X state weights corresponding one-to-one to X states, step 602 may be implemented by determining the target sensor configuration based on the X state weights.

[0169] For example, in a scenario where multiple candidate sensor configurations are pre-configured, determining the target sensor configuration based on X state weights may be implemented by determining the target sensor configuration from the multiple candidate sensor configurations based on the X state weights.

[0170] For example, X state weights are input to a classification model so that the classification model outputs a target sensor configuration.

[0171] The above-mentioned multiple candidate sensor configurations and classification models are all pre-trained, which is explained below.

[0172] 1. Training multiple candidate sensor configurations

[0173] In some embodiments, training to obtain multiple candidate sensor configurations may be implemented by: determining multiple initial sensor configurations, each of the multiple initial sensor configurations including the operating frequency of the sensor of each of the M movable objects; determining a safety level of each of the multiple initial sensor configurations, the safety level indicating the possibility that the simulated formation can travel safely after being set up according to the initial sensor configuration; and determining multiple candidate sensor configurations from the multiple initial sensor configurations based on the safety level of each of the multiple initial sensor configurations.

[0174] That is, a plurality of candidate sensor configurations are screened out from a plurality of initial sensor configurations based on the possibility of safe driving of the simulated platoon.

[0175] The simulated formation can be a virtual formation simulated by a computer. In this scenario, the safety level of each initial sensor configuration is determined by the likelihood that the simulated formation will safely navigate the simulated virtual scenario. This method allows for rapid training of multiple candidate sensor configurations.

[0176] Based on this, the implementation method for determining the safety level of each initial sensor configuration among multiple initial sensor configurations can be: for any initial sensor configuration among the multiple initial sensor configurations, setting the sensors of each simulated object in the simulated formation according to the initial sensor configuration, and making the simulated formation travel in the scenario use case of each simulated scene in multiple simulated scenes; based on the driving conditions of the simulated formation in the scenario use case of each simulated scene, determining the number of scenario use cases in which the simulated formation can travel safely in the multiple simulated scenes; and determining the ratio between the number of scenario use cases that can travel safely and the total number of scenario use cases in the multiple simulated scenes as the safety level of the initial sensor configuration.

[0177] Figure 7 is a schematic diagram of a simulation scenario provided by an embodiment of the present application. As shown in Figure 7, the simulation scenario may include a straight-ahead scenario, a turning scenario, a roundabout scenario, a lane-changing scenario, a blocked scenario, etc.

[0178] Because the surrounding environment of the formation can vary in various ways in each simulation scenario, and the tasks performed by each movable object in the formation can also vary, in order to accurately measure the safety level of each initial sensor configuration, a large number of scenario use cases can be set for each simulation scenario. Different scenario use cases have different surrounding environments and / or different task sequences performed by the same movable object in the formation. For example, for the straight-ahead scenario in Figure 7, 500 scenario use cases can be set, some of which have different positions and / or numbers of pedestrians around the formation, some of which have different lanes, and some of which have different positions and / or numbers of vehicles around the formation.

[0179] Among them, the implementation method of judging whether the simulated formation is driving safely in a certain scenario use case can be: if the simulated formation does not violate traffic rules in the scenario use case, then it is determined that the simulated formation can drive safely in the scenario use case, wherein examples of events that violate traffic rules include speeding events, collision events, red light running events, etc.

[0180] Furthermore, in the aforementioned implementation of determining whether the simulated formation is operating safely in a particular scenario, in addition to considering whether the simulated formation violates traffic regulations in that scenario, it may also be considered whether the simulated formation exits the intelligent driving function during the operation of that scenario. If the simulated formation does not violate traffic regulations in that scenario and the simulated formation does not exit the intelligent driving function throughout the entire operation of that scenario, the simulated formation is determined to be operating safely in that scenario.

[0181] FIG8 is a flow chart of a training process for obtaining multiple candidate sensor configurations according to an embodiment of the present application. Assuming that the number of candidate sensor configurations to be trained is 5, as shown in FIG8 , the training process is as follows:

[0182] 1. Set i = 1.

[0183] 2. First determine whether i is less than or equal to 5. If the result is yes, set score = 0 and W = 0.

[0184] 3. Generate an initial sensor configuration in a random manner. The generation method may be: for a sensor on a movable object, generate a random frequency of the sensor using the following formula, and use the random frequency as the operating frequency of the sensor in the generated initial sensor configuration.

[0185]

[0186] Among them, f jmin is the minimum supported frequency of the sensor (can be 0), f jmax is the maximum supported frequency of the sensor, and the value in the above formula [] is the range of random number generation.

[0187] A frequency is randomly generated for each configurable sensor in the formation using the above formula until a complete initial sensor configuration a is generated.

[0188] 4. Input the initial sensor configuration a into the simulator to adjust the sensor configuration of the simulated vehicle according to the initial sensor configuration a, and run scenario use cases in different simulation scenarios. The scenario use cases are preset by the system.

[0189] 5. Run all scenario use cases to determine the safety level of initial sensor configuration a (i.e., the percentage of scenario use cases that pass). If the safety level exceeds the score, initial sensor configuration a is assigned to the candidate sensor configuration, and the determined safety level is assigned to the score. Conversely, if the safety level does not exceed the score, initial sensor configuration a is not selected as a candidate, and the score is not updated. Next, determine whether W is less than 1000.

[0190] 6. If W is less than 1000, return to step 3 and repeat steps 3-6 until W = 1000. The resulting candidate sensor configuration is then used as the i-th candidate sensor configuration. After 1000 cycles, the initial sensor configuration a with the highest security level is obtained. Therefore, this initial sensor configuration a with the highest security level is used as the i-th candidate sensor configuration.

[0191] 7. When W reaches 1000, increase the value of i by 1 and return to step 2 until 5 candidate sensor configurations are generated.

[0192] Through steps 1-7 above, a candidate sensor configuration is obtained in each of the five cycles. In each cycle, 1,000 initial sensor configurations are generated and fed into the simulator. The initial sensor configuration with the highest security level is selected as the candidate sensor configuration for training.

[0193] Figure 8 illustrates an example method for training multiple candidate sensor configurations. Alternatively, steps 3-5 can be performed directly to obtain the security level of each of the 1,000 initial sensor configurations, and then the top five initial sensor configurations with the highest security levels can be directly selected as the five candidate sensor configurations obtained through training.

[0194] Alternatively, the simulated formation can be a formation other than the one in step 601 that has already completed its driving mission. In this scenario, the safety level of each initial sensor configuration can be determined by the simulated formation's ability to safely drive in various real-world scenarios before the current time. In this way, multiple candidate sensor configurations can be accurately trained.

[0195] After training multiple candidate sensor configurations, you can also set the security level of each candidate sensor configuration. For example, after obtaining five candidate sensor configurations, they are sorted in descending order of security. The security level of the first candidate sensor configuration is set to 1, the security level of the second candidate sensor configuration is set to 2, and so on, and the security level of the last candidate sensor configuration is set to 5. This means that security level 1 is the highest security level, and security level 5 is the lowest security level.

[0196] In some scenarios, all sensors in the candidate sensor configuration corresponding to security level 1 are turned on, and all configurable sensors in the candidate sensor configuration corresponding to security level 5 are turned off or run at a low frequency.

[0197] In addition, after multiple candidate sensor configurations are obtained through training, the multiple candidate sensor configurations can be written into a sensor configuration file for subsequent use.

[0198] In some embodiments, the pseudo code (json format) of the sensor configuration file can be expressed as follows. The following code is explained using the code of one sensor in the sensor configuration file as an example:

[0199] "Front_Radar":{"car_num":"A","Status":"disable","frenquency":"10"}

[0200] Here, "Front_Radar" represents the sensor name.

[0201] “car_num”: “1” represents the car with vehicle number A in the formation.

[0202] "Status": "disable" means that the vehicle's front radar sensor status needs to be set to off.

[0203] "frequency": "10" means that when Status is enable, the sensor runs at a frequency of 10 Hz.

[0204] 2. Training the Classification Model

[0205] In some embodiments, the classification model includes a minimum regression decision tree. In this scenario, the classification model can be trained by: obtaining multiple sample state data, each sample state data indicating a state of the simulated formation; constructing an initialized minimum regression decision tree based on the multiple sample state data, the initialized minimum regression decision tree including multiple initial classification results, the number of the multiple initial classification results being the same as the number of the multiple candidate sensor configurations, and different initial classification results indicating different levels of state complexity of the simulated formation; binding the multiple initial classification results to the multiple candidate sensor configurations one by one to obtain a minimum regression decision tree; wherein, the greater the level of state complexity indicated by an initial classification result among the multiple initial classification results, the higher the sensor operating frequency in the candidate sensor configuration bound to the corresponding initial classification result.

[0206] Each sample state data may include multiple sample state weights. For example, each sample state data may include an environment state weight, a task state weight, and a load state weight.

[0207] In some embodiments, a plurality of sample state data may be obtained from an analysis of the driving process of the aforementioned simulated formation in each scenario use case of each of a plurality of simulated scenarios.

[0208] In addition, constructing an initialized minimum regression decision tree based on multiple sample state data is also called a tree building process. In some embodiments, the tree building process can be implemented through the following steps.

[0209] a) Obtain multiple sample state data including three features: environment state weight, task state weight, and load state weight. For ease of explanation, the environment state weight is labeled as feature 1, the task state weight is labeled as feature 2, and the load state weight is labeled as feature 3. These multiple sample state data can be represented by the following data set D:

[0210] D=(x1,y1,z1), (x2,y2,z2), (x3,y3,z3),…, (xm,ym,zm).

[0211] b) For feature 1, first randomly obtain xj and determine the segmentation space R(j, s) corresponding to xj, where j is the subscript of x and s is the actual sample value of xj.

[0212] R1(j, s) = {x|xj≤s} and R2(j, s) = {x|xj>s}

[0213] c) Solve the optimal split variable j and split point s using the following formula:

[0214] min j,s[min c1∑xi∈R1(j,s)(yi-c1) / 2+min c2∑xi∈R2(j,s)(yi-c2) / 2]

[0215] Where c1 and c2 are the means of the samples of feature 1 falling into the first segmentation space and the second switching space, respectively.

[0216] Specifically, we change the value of j to traverse all samples of feature 1. Using the above formula, we find the optimal split variable j, forming a pair (j, s). This pair then divides the numerical space corresponding to feature 1 into two regions. We then repeat this division process for each region until the conditions are met. Ultimately, the numerical space corresponding to feature 1 is divided into y regions, denoted by R1, R2, …, Ry, where y is greater than or equal to 2. This forms the first layer of the decision tree.

[0217] Because the sample data contains three features, assume that feature 1 is split into three leaf nodes, c11, c12, and c13, as shown in Figure 9. These three leaf nodes correspond to the three spaces after feature 1 is split. Then, based on the three leaf nodes c11, c12, and c13 after feature 1 is split, feature 2 is split. As shown in Figure 9, assume that feature 2 is split into two leaf nodes, c21 and c22, which correspond to the two spaces after feature 2 is split. Then, based on the space after feature 2 is split, feature 3 is split. As shown in Figure 9, assume that feature 3 is split into five leaf nodes, c31, c32, c33, c34, and c35, which correspond to the five spaces after feature 3 is split. In this way, the initialized minimum regression decision tree with a depth of 3 is constructed, as shown in Figure 9. The five leaf nodes c31, c32, c33, c34, and c35 represent the five initial classification results.

[0218] As shown in FIG9 , after obtaining the initialized minimum regression decision tree, it is necessary to bind the five initial classification results of the initialized minimum regression decision tree and the five candidate sensor configurations obtained from the aforementioned training respectively, so as to use the candidate sensor configuration as the final classification result.

[0219] As shown in FIG9 , the initial classification result c31 is bound to the candidate sensor configuration 1, the initial classification result c32 is bound to the candidate sensor configuration 2, the initial classification result c33 is bound to the candidate sensor configuration 3, the initial classification result c34 is bound to the candidate sensor configuration 4, and the initial classification result c35 is bound to the candidate sensor configuration 5.

[0220] Different initial classification results indicate different levels of state complexity for the formation. Based on this, when binding the initial classification results to candidate sensor configurations, the greater the state complexity level indicated by an initial classification result among multiple initial classification results, the higher the sensor operating frequency in the candidate sensor configuration bound to the corresponding initial classification result, that is, the higher the security level of the bound initial sensor configuration.

[0221] For example, in Figure 9, initial classification result c31 indicates that the surrounding environment, the status of the leader's task, and the load status of each movable object are all very complex. Therefore, initial classification result c31 is bound to candidate sensor configuration 1, which has the highest security level. Initial classification result c35 indicates that the surrounding environment, the status of the leader's task, and the load status of each movable object are not very complex. Therefore, initial classification result c35 is bound to candidate sensor configuration 5, which has the lowest security level.

[0222] In addition, as shown in FIG9 , the generation process of the minimum regression decision tree can be implemented by the classification model generation module shown in FIG4 or FIG5 .

[0223] Based on the minimum regression decision tree shown in Figure 9, when the current state data of the formation is obtained through step 601, and the current state data includes the environment state weight w1, the task state weight w2, and the load state weight w3, these three state weights are input into the minimum regression decision tree, and the minimum regression decision tree will output a candidate sensor configuration, that is, the target sensor configuration.

[0224] The above minimum regression decision tree is an example of a classification model. Alternatively, other neural network models can be used to classify the current state data into one of the candidate sensor configurations, which will not be described in detail here.

[0225] Step 603: Send corresponding sensor operating frequency information to each of the N movable objects, so that the movable object adjusts the operating frequency of the sensor on the movable object based on the corresponding sensor operating frequency information.

[0226] Since the target sensor configuration includes the sensor operating frequency information of N movable objects among the M movable objects in the formation, after obtaining the target sensor configuration in step 602, the sensor operating frequency information of each movable object among the N movable objects is sent to the movable object so that the movable object adjusts the operating frequency of its own sensor.

[0227] In some embodiments, for any one of the N movable objects, a sensor frequency adjustment instruction may be generated based on the sensor operating frequency information of the movable object in the target sensor configuration, and the sensor frequency adjustment instruction may be issued to the movable object. Upon receiving the sensor frequency adjustment instruction, the movable object may adjust the operating frequency of the corresponding sensor based on the sensor frequency adjustment instruction.

[0228] For example, a sensor frequency adjustment instruction could be: "Change the frequency of the rear telephoto camera of Team Leader Car A from 30 Hz to 10 Hz." This frequency adjustment instruction is sent to Team Leader Car A's electronic control unit (ECU) or domain controller via the communication network, causing Team Leader Car A to control the frequency of its rear telephoto camera from 30 Hz to 10 Hz.

[0229] In addition, a sensor frequency adjustment instruction may be generated for each configurable sensor on the movable object, so that multiple sensor frequency adjustment instructions may be generated. These multiple sensor frequency adjustment instructions may be sent to the movable object as an instruction sequence.

[0230] Furthermore, in some embodiments, to prevent safety issues caused by the loss of some sensor information during the sensor operating frequency switching process, after receiving a sensor frequency adjustment instruction, the movable object can pre-store sensor data for a period of time (T = 10 seconds) for use in intelligent driving tasks during the sensor switching process. After the pre-storage is complete, the sensor is controlled to adjust its operating frequency based on the sensor frequency adjustment instruction. The sensor status is then checked to see if it has changed accordingly. If it has not changed according to the sensor frequency adjustment instruction, a sensor failure log is reported; otherwise, a frequency adjustment success log is reported.

[0231] In addition, when the movable object adjusts the operating frequency of a certain sensor of its own according to the sensor frequency adjustment instruction, the movable object may also adjust the algorithm corresponding to the sensor. The algorithm corresponding to the sensor may, for example, include a related algorithm deployed on a controller configured on the sensor. For example, for a camera, the algorithm corresponding to the camera may include a target detection algorithm deployed on a controller configured on the camera. The algorithm corresponding to the sensor may, for example, also include an algorithm for the speed at which the movable object reads data from the sensor, wherein the algorithm for the speed at which the movable object reads data from the sensor may also be referred to as a callback function. The examples for adjusting the algorithm corresponding to the sensor will not be given one by one here.

[0232] For example, if the sensor's operating frequency is adjusted to 0, meaning the sensor is turned off, the algorithm corresponding to the sensor can be turned off accordingly. In this case, the sensor can be marked as "standby" in the program. For another example, if the sensor's operating frequency is adjusted from 0 to a certain value, meaning the sensor is turned on, the algorithm corresponding to the sensor can be turned on. In this case, the sensor's flag can be switched from "standby" to "active" in the program. For another example, if the sensor's operating frequency decreases, the frequency of reading data in the callback function can be reduced accordingly.

[0233] After the movable object has adjusted the operating frequency of the sensor and the corresponding algorithm, it will restart acquiring the current state data after a certain period of time (for example, 3 minutes) to perform the next adjustment process.

[0234] In summary, in an embodiment of the present application, the operating frequency of the sensors on the movable objects in the formation can be adjusted based on the current status data of the formation, so as to flexibly adjust the power consumption of the formation based on the current status data, thereby avoiding the problem of excessive power consumption of the formation in certain scenarios.

[0235] Based on the formation control system shown in FIG4 , it can be seen that the embodiment of the present application can also switch the roles in the formation to further reduce the power consumption of the formation. This is explained below.

[0236] FIG10 is a flow chart of a process of switching formation roles provided by an embodiment of the present application. As shown in FIG10 , the method includes the following steps 1001 to 1003 .

[0237] Step 1001: Acquire current status information of a sensor of each of the M movable objects in the formation.

[0238] The current state of the sensor may include information such as the power, operating frequency, and temperature of the sensor.

[0239] Step 1002: Determine the overall energy efficiency ratio of the formation based on current status information of a sensor of each of the M movable objects.

[0240] The overall energy efficiency ratio (EER) indicates the overall energy utilization of the formation. A higher EER indicates a higher overall energy utilization rate for the formation. A lower EER indicates a lower overall energy utilization rate for the formation. In this case, the EER can be improved by adjusting the formation roles, thereby reducing power consumption.

[0241] In some embodiments, the overall energy efficiency ratio of the formation can be determined by the following formula:

[0242]

[0243] Among them, W is the overall energy efficiency ratio of the formation, Xi is the energy efficiency ratio of a sensor on a movable object in the formation, and wi is the weight of the sensor.

[0244] Xi can be obtained from the frequency, temperature, power and other information of the sensor and the power characteristic curve of the sensor. wi is a preset weight, which can be set according to the importance of the sensor in the formation.

[0245] The above formula is used to illustrate how to determine the overall energy efficiency ratio of a formation. Alternatively, the overall energy efficiency ratio of a formation can be determined by other methods, which will not be described here one by one.

[0246] Step 1003: If the overall energy efficiency ratio is lower than the reference energy efficiency ratio, the roles of at least two movable objects in the formation are switched.

[0247] The reference energy efficiency ratio is a preset value. If the overall energy efficiency ratio is lower than the reference energy efficiency ratio, it indicates that the current power consumption of the formation is relatively high. For example, the reference energy efficiency ratio may be 60%.

[0248] Among them, switching the roles of at least two movable objects in the formation may include: adjusting an original follower to a leader, adjusting the original leader to a follower, or exchanging the positions of the original two followers in the lane, etc.

[0249] Figure 11 is a schematic diagram of a scenario for switching formation roles, provided in an embodiment of the present application. As shown in Figure 11, vehicles A, B, and C form a formation. Before the adjustment, vehicle A is the leader, and vehicles B and C are followers. After the adjustment, vehicle B is the leader, and vehicles A and C are followers.

[0250] In some embodiments, if the overall energy efficiency ratio is lower than a reference energy efficiency ratio, the roles of the at least two movable objects in the formation may be directly switched. Optionally, if the overall energy efficiency ratio is lower than the reference energy efficiency ratio, the expected power consumption of the switching process may be further determined, where the switching process is the process of switching the roles of the at least two movable objects in the formation; the power difference between the power before and after the switching of the formation is determined; and if the power difference exceeds the expected power consumption, the roles of the at least two movable objects in the formation are switched.

[0251] If the power difference exceeds the expected power consumption, it indicates that the switching process is relatively inexpensive, and in this case, the roles of the at least two movable objects in the formation can be switched. Correspondingly, if the power difference is lower than the expected power consumption, it indicates that the switching process is relatively expensive, and in this case, the roles of the at least two movable objects in the formation can be switched, thereby avoiding the need for further power consumption.

[0252] In some embodiments, determining the expected power consumption of the handover process may be achieved by: determining at least one task required to be executed during the handover process; and determining the expected power consumption based on the empirical power of each of the at least one task, where the empirical power indicates the power required when the corresponding task was executed before the current time. The empirical power may also be referred to as an empirical value of the power consumption of the task.

[0253] For example, the actual power of each task before the current time can be counted in different tasks such as side collision detection, collaborative path planning, control sequence execution, etc. For any task, the distribution of the actual power of the task can be determined based on the multiple execution processes of the task before the current time. The distribution of the actual power of the task usually presents a normal distribution (which can be expressed by μσ 2 The actual average power of the task can be determined by the distribution of the actual power of the task, and the actual average power can be used as the empirical power of the task.

[0254] Figure 12 is a schematic diagram of another process flow for switching formation roles provided by an embodiment of the present application. As shown in Figure 12, after the configuration of each sensor in the embodiment of Figure 6 is successful, the current state information (such as temperature / frequency / power) of each sensor of each movable object can be extracted. Based on the current state information of each sensor of each movable object and the sensor power characteristic curve, the overall energy efficiency ratio of the formation is determined. It is determined whether the overall energy efficiency ratio of the formation is less than 60%. If the overall energy efficiency ratio of the formation exceeds 60%, the operation of extracting the current state information of each sensor of each movable object is re-executed after sleeping for 30 seconds.

[0255] If the overall energy efficiency ratio of the formation is lower than 60%, the expected power consumption P1 of the switching process and the power difference P2 between the power before and after the switching of the formation are determined. If P1 is less than P2, dynamic adjustment is initiated to adjust the roles of at least two movable objects in the formation.

[0256] In summary, in the embodiment of the present application, the roles of the movable objects in the formation can also be adjusted according to the energy efficiency ratio of the formation to further reduce the power consumption of the formation.

[0257] In addition, as shown in Figure 13 , an embodiment of the present application further provides a formation control device 1300, which includes a processing unit 1301 and a communication unit 1302. The various modules in the formation control system shown in Figures 4 and 5 can be deployed in the processing unit 1301 or the communication unit 1302 to implement the formation control method provided in the embodiment of the present application.

[0258] The processing unit 1301 is configured to obtain current state data of a formation, where the formation includes M movable objects, where M is a positive integer greater than or equal to 2. For a specific implementation, reference may be made to step 601 in the embodiment of FIG6 .

[0259] The processing unit 1302 is further configured to determine a target sensor configuration based on the current state data. The target sensor configuration includes sensor operating frequency information for N of the M movable objects, and the target sensor configuration is capable of achieving safe platooning, where N is less than or equal to M. For a specific implementation, see step 602 in the embodiment of FIG. 6 .

[0260] The communication unit 1302 is configured to send corresponding sensor operating frequency information to each of the N movable objects, so that the movable object adjusts the operating frequency of the sensor on the movable object based on the corresponding sensor operating frequency information. For specific implementation, see step 603 in the embodiment of FIG. 6 .

[0261] Optionally, when determining the target sensor configuration based on the current state data, the processing unit 1302 is configured to:

[0262] A target sensor configuration is determined from a plurality of candidate sensor configurations based on the current state data.

[0263] Optionally, the processing unit 1302 is further configured to:

[0264] determining a plurality of initial sensor configurations, each of the plurality of initial sensor configurations including an operating frequency of a sensor of each of the M movable objects;

[0265] determining a safety level for each of the plurality of initial sensor configurations, the safety level indicating a likelihood that the simulated formation can safely travel according to the initial sensor configuration;

[0266] A plurality of candidate sensor configurations are determined from the plurality of initial sensor configurations based on a security level of each of the plurality of initial sensor configurations.

[0267] Optionally, when determining the security level of each of the multiple initial sensor configurations, the processing unit 1302 is configured to:

[0268] For any initial sensor configuration among the multiple initial sensor configurations, setting a sensor of each simulated object in the simulated formation according to the initial sensor configuration, and causing the simulated formation to travel in a scenario use case of each simulated scenario among the multiple simulation scenarios;

[0269] Determining the number of scenario use cases in which the simulated formation can safely travel in the multiple simulation scenarios based on the driving conditions of the simulated formation in the scenario use cases of each simulation scenario;

[0270] The ratio between the number of scenario use cases that can be driven safely and the total number of scenario use cases of multiple simulation scenarios is determined as the safety level of any initial sensor configuration.

[0271] Optionally, when determining the target sensor configuration from a plurality of candidate sensor configurations based on the current state data, the processing unit 1302 is configured to:

[0272] inputting the current state data into the classification model so that the classification model outputs the target sensor configuration;

[0273] The classification model is used to classify the input data into one of multiple candidate sensor configurations and output the classification result.

[0274] Optionally, the classification model includes a minimum regression decision tree; the processing unit 1302 is further configured to:

[0275] Acquire a plurality of sample state data, each sample state data indicating a state of the simulated formation;

[0276] constructing an initialized minimum regression decision tree based on multiple sample state data, wherein the initialized minimum regression decision tree includes multiple initial classification results, the number of the multiple initial classification results is the same as the number of the multiple candidate sensor configurations, and different initial classification results indicate different state complexities of the simulated formation;

[0277] Bind multiple initial classification results to multiple candidate sensor configurations one by one to obtain a minimum regression decision tree;

[0278] The greater the state complexity level indicated by an initial classification result among the multiple initial classification results, the higher the sensor operating frequency in the candidate sensor configuration bound to the corresponding initial classification result.

[0279] Optionally, the current state data includes X state weights corresponding to the X states one by one, each state weight in the X state weights indicates the complexity of the corresponding state, and X is greater than or equal to 1;

[0280] When determining the target sensor configuration based on the current state data, the processing unit 1302 is configured to:

[0281] A target sensor configuration is determined based on the X state weights.

[0282] Optionally, the X states include one or more of an environment state, a task state, and a load state;

[0283] Among them, the state weight corresponding to the environmental state includes the environmental state weight of at least one movable object in the formation, and the environmental state weight indicates the complexity of the environmental state around the corresponding movable object. The state weight corresponding to the task state indicates the complexity of the task currently performed by the leader in the formation. The state weight corresponding to the load state includes the load state weight of each movable object in the formation, and the load state weight indicates the complexity of the load of the corresponding movable object.

[0284] Optionally, the processing unit 1302 is further configured to:

[0285] The current state information of the sensor of each of the M movable objects is obtained. For specific implementation, please refer to step 1001 in the embodiment of FIG10 .

[0286] The overall energy efficiency ratio of the formation is determined based on the current state information of the sensor of each of the M movable objects. For specific implementation, please refer to step 1002 in the embodiment of Figure 10.

[0287] If the overall energy efficiency ratio is lower than the reference energy efficiency ratio, the roles of at least two movable objects in the formation are switched. For a specific implementation, reference may be made to step 1003 in the embodiment of FIG. 10 .

[0288] Optionally, if the overall energy efficiency ratio is lower than the reference energy efficiency ratio, the processing unit 1302 switches the roles of at least two movable objects in the formation, to:

[0289] If the overall energy efficiency ratio is lower than the reference energy efficiency ratio, determining an expected power consumption of a switching process, where the switching process is a process of switching roles of at least two movable objects in the formation;

[0290] Determine the power difference between the power before and after the switching of the formation;

[0291] If the power difference exceeds the expected consumed power, the roles of at least two movable objects in the formation are switched.

[0292] Optionally, when determining the expected power consumption of the handover process, the processing unit 1302 is configured to:

[0293] determining at least one task to be performed during the switching process;

[0294] An expected consumed power is determined based on an experienced power of each of the at least one task, the experienced power indicating a power required when the corresponding task was executed before a current time.

[0295] In embodiments of the present application, the operating frequency of sensors on movable objects within the formation can be adjusted based on the formation's current state data. This allows for flexible adjustment of the formation's power consumption based on the current state data, thereby avoiding excessive power consumption in certain scenarios. Furthermore, by adjusting the operating frequency of sensors within the formation, such as by lowering it, the amount of data collected by the sensors will decrease, and accordingly, the amount of data the formation needs to process will also decrease, thereby conserving the formation's computing resources.

[0296] It should be noted that the above-described embodiments of the formation control device, when controlling a formation, illustrate the division of the aforementioned functional units only as an example. In actual applications, the aforementioned functions can be assigned to different functional units as needed, i.e., the internal structure of the device can be divided into different functional units to perform all or part of the functions described above. Furthermore, the formation control device and the formation control method embodiments provided in the above-described embodiments share the same concept. The specific implementation process is detailed in the method embodiments and will not be further elaborated here.

[0297] Those skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware, or by a program to instruct the relevant hardware, and the program may be stored in a computer-readable storage medium, which may be a read-only memory, a disk, or an optical disk, etc.

[0298] The above description is merely an exemplary embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A formation control method, characterized in that: The method comprises: Obtaining current state data of a formation, where the formation includes M movable objects, where M is a positive integer greater than or equal to 2; determining a target sensor configuration based on the current state data, the target sensor configuration including sensor operating frequency information of N movable objects among the M movable objects, and the target sensor configuration enabling safe driving of the formation, where N is less than or equal to M; Corresponding sensor operating frequency information is sent to each of the N movable objects, so that the movable object adjusts the operating frequency of the sensor on the movable object based on the corresponding sensor operating frequency information.

2. The method according to claim 1, wherein The determining of the target sensor configuration based on the current state data includes: The target sensor configuration is determined from a plurality of candidate sensor configurations based on the current state data.

3. The method according to claim 2, wherein The method further comprises: determining a plurality of initial sensor configurations, each of the plurality of initial sensor configurations comprising an operating frequency of a sensor of each of the M movable objects; determining a safety level for each of the plurality of initial sensor configurations, the safety level indicating a likelihood that the simulated formation can safely travel according to the initial sensor configuration; The plurality of candidate sensor configurations are determined from the plurality of initial sensor configurations based on a security level of each of the plurality of initial sensor configurations.

4. The method according to claim 3, wherein Determining the security level of each initial sensor configuration in the plurality of initial sensor configurations includes: For any initial sensor configuration among the multiple initial sensor configurations, setting a sensor of each simulated object in the simulated formation according to the any initial sensor configuration, and causing the simulated formation to travel in a scenario use case of each simulated scenario among the multiple simulated scenarios; Determining, based on the driving condition of the simulated formation in the scenario use case of each simulated scenario, the number of scenario use cases in which the simulated formation can safely drive in the multiple simulated scenarios; The ratio between the number of scenario use cases that enable safe driving and the total number of scenario use cases of the multiple simulation scenarios is determined as the safety level of any one of the initial sensor configurations.

5. The method according to any one of claims 2 to 4, characterized in that: The determining the target sensor configuration from a plurality of candidate sensor configurations based on the current state data comprises: inputting the current state data into a classification model so that the classification model outputs the target sensor configuration; The classification model is used to classify input data into one of the plurality of candidate sensor configurations and output a classification result.

6. The method of claim 5, wherein the classification model comprises a minimum regression decision tree, and the method further comprises: Acquire a plurality of sample state data, each sample state data indicating a state of the simulated formation; constructing an initialized minimum regression decision tree based on the multiple sample state data, the initialized minimum regression decision tree including multiple initial classification results, the number of the multiple initial classification results being the same as the number of the multiple candidate sensor configurations, and different initial classification results indicating different state complexities of the simulated formation; Binding the multiple initial classification results to the multiple candidate sensor configurations one by one to obtain the minimum regression decision tree; The greater the state complexity level indicated by an initial classification result among the multiple initial classification results, the higher the sensor operating frequency in the candidate sensor configuration bound to the corresponding initial classification result.

7. The method according to any one of claims 1 to 6, wherein: The current state data includes X state weights corresponding to the X states one by one, each state weight in the X state weights indicates the complexity of the corresponding state, and X is greater than or equal to 1; The determining of the target sensor configuration based on the current state data includes: The target sensor configuration is determined based on the X state weights.

8. The method according to claim 7, wherein The X states include one or more of an environment state, a task state, and a load state; Among them, the state weight corresponding to the environmental state includes the environmental state weight of at least one movable object in the formation, and the environmental state weight indicates the complexity of the environmental state around the corresponding movable object. The state weight corresponding to the task state indicates the complexity of the task currently performed by the leader in the formation. The state weight corresponding to the load state includes the load state weight of each movable object in the formation, and the load state weight indicates the complexity of the load of the corresponding movable object.

9. The method according to any one of claims 1 to 8, wherein: The method further comprises: Acquire current state information of a sensor of each of the M movable objects; determining an overall energy efficiency ratio of the formation based on current state information of a sensor of each of the M movable objects; If the overall energy efficiency ratio is lower than a reference energy efficiency ratio, roles of at least two movable objects in the formation are switched.

10. The method according to claim 9, wherein If the overall energy efficiency ratio is lower than a reference energy efficiency ratio, switching roles of at least two movable objects in the formation includes: If the overall energy efficiency ratio is lower than a reference energy efficiency ratio, determining an expected power consumption of a switching process, wherein the switching process is a process of switching roles of at least two movable objects in the formation; Determining a power difference between a power before switching and a power after switching of the formation; If the power difference exceeds the expected consumed power, roles of at least two movable objects in the formation are switched.

11. The method according to claim 10, wherein The determining of the expected power consumption of the switching process includes: Determining at least one task to be performed during the handover process; The expected consumed power is determined based on an empirical power of each of the at least one task, the empirical power indicating power required when the corresponding task was executed before a current time.

12. A formation control device, characterized in that: The device comprises: a processing unit, configured to obtain current state data of a formation, wherein the formation includes M movable objects, where M is a positive integer greater than or equal to 2; The processing unit is further configured to determine a target sensor configuration based on the current state data, the target sensor configuration including sensor operating frequency information of N movable objects among the M movable objects, and the target sensor configuration is capable of achieving safe driving of the formation, where N is less than or equal to M; The communication unit is configured to send corresponding sensor operating frequency information to each of the N movable objects, so that the movable object can adjust the operating frequency of the sensor on the movable object based on the corresponding sensor operating frequency information.

13. The device according to claim 12, wherein When determining the target sensor configuration based on the current state data, the processing unit is configured to: The target sensor configuration is determined from a plurality of candidate sensor configurations based on the current state data.

14. The device according to claim 13, wherein The processing unit is further configured to: determining a plurality of initial sensor configurations, each of the plurality of initial sensor configurations comprising an operating frequency of a sensor of each of the M movable objects; determining a safety level for each of the plurality of initial sensor configurations, the safety level indicating a likelihood that the simulated formation can safely travel according to the initial sensor configuration; The plurality of candidate sensor configurations are determined from the plurality of initial sensor configurations based on a security level of each of the plurality of initial sensor configurations.

15. The device according to claim 14, wherein The processing unit, when determining the security level of each initial sensor configuration in the plurality of initial sensor configurations, is configured to: For any initial sensor configuration among the multiple initial sensor configurations, setting a sensor of each simulated object in the simulated formation according to the any initial sensor configuration, and causing the simulated formation to travel in a scenario use case of each simulated scenario among the multiple simulated scenarios; Determining, based on the driving condition of the simulated formation in the scenario use case of each simulated scenario, the number of scenario use cases in which the simulated formation can safely drive in the multiple simulated scenarios; The ratio between the number of scenario use cases that enable safe driving and the total number of scenario use cases of the multiple simulation scenarios is determined as the safety level of any one of the initial sensor configurations.

16. The device according to any one of claims 13 to 15, characterized in that: When determining the target sensor configuration from a plurality of candidate sensor configurations based on the current state data, the processing unit is configured to: inputting the current state data into a classification model so that the classification model outputs the target sensor configuration; The classification model is used to classify input data into one of the plurality of candidate sensor configurations and output a classification result.

17. The apparatus according to claim 16, wherein the classification model comprises a minimum regression decision tree, and the processing unit is further configured to: Acquire a plurality of sample state data, each sample state data indicating a state of the simulated formation; constructing an initialized minimum regression decision tree based on the multiple sample state data, the initialized minimum regression decision tree including multiple initial classification results, the number of the multiple initial classification results being the same as the number of the multiple candidate sensor configurations, and different initial classification results indicating different state complexities of the simulated formation; Binding the multiple initial classification results to the multiple candidate sensor configurations one by one to obtain the minimum regression decision tree; in, The greater the state complexity level indicated by an initial classification result among the multiple initial classification results, the higher the sensor operating frequency in the candidate sensor configuration bound to the corresponding initial classification result.

18. The device according to any one of claims 12 to 17, characterized in that: The current state data includes X state weights corresponding to the X states one by one, each state weight in the X state weights indicates the complexity of the corresponding state, and X is greater than or equal to 1; When determining the target sensor configuration based on the current state data, the processing unit is configured to: The target sensor configuration is determined based on the X state weights.

19. The device according to claim 18, wherein The X states include one or more of an environment state, a task state, and a load state; Among them, the state weight corresponding to the environmental state includes the environmental state weight of at least one movable object in the formation, and the environmental state weight indicates the complexity of the environmental state around the corresponding movable object. The state weight corresponding to the task state indicates the complexity of the task currently performed by the leader in the formation. The state weight corresponding to the load state includes the load state weight of each movable object in the formation, and the load state weight indicates the complexity of the load of the corresponding movable object.

20. The device according to any one of claims 12 to 19, characterized in that: The processing unit is further configured to: Acquire current state information of a sensor of each of the M movable objects; determining an overall energy efficiency ratio of the formation based on current state information of a sensor of each of the M movable objects; If the overall energy efficiency ratio is lower than a reference energy efficiency ratio, roles of at least two movable objects in the formation are switched.

21. The device according to claim 20, characterized in that The processing unit switches roles of at least two movable objects in the formation if the overall energy efficiency ratio is lower than a reference energy efficiency ratio, and is configured to: If the overall energy efficiency ratio is lower than a reference energy efficiency ratio, determining an expected power consumption of a switching process, wherein the switching process is a process of switching roles of at least two movable objects in the formation; Determining a power difference between a power before switching and a power after switching of the formation; If the power difference exceeds the expected consumed power, roles of at least two movable objects in the formation are switched.

22. The device according to claim 21, wherein When determining the expected power consumption of the switching process, the processing unit is configured to: Determining at least one task to be performed during the handover process; The expected consumed power is determined based on an empirical power of each of the at least one task, the empirical power indicating power required when the corresponding task was executed before a current time.

23. A formation control device, characterized in that: including processor and memory; The memory is used to store a program that supports the formation control device to execute the method according to any one of claims 1 to 11, and to store data involved in implementing the method according to any one of claims 1 to 11; The processor is configured to execute the program stored in the memory.

24. An intelligent driving device, characterized in that: Comprising the formation control device as described in any one of claims 12-23.

Images