CN116783461A - Vehicle testing methods and systems based on unmanned aerial vehicles - Google Patents

Abstract

A vehicle testing method and system based on unmanned aerial vehicles. The method includes: sending preset test instructions to the unmanned aerial vehicle. The test instructions are used to instruct the unmanned aerial vehicle to move in a traffic scene, wherein the unmanned aerial vehicle carries a target object. The model of the target object moves with the UAV in the traffic scene (S110); obtains the driving state of the vehicle under test in the traffic scene (S120); generates the test results of the vehicle under test according to the driving state of the vehicle under test (S120) S130). The target model has better maneuverability and sensitivity.

Description

Vehicle testing methods and systems based on unmanned aerial vehicles Technical field

This application relates to the field of autonomous driving technology, and in particular to a vehicle testing method and system based on unmanned aerial vehicles.

Background technique

Autonomous driving means that the driver does not need to operate the vehicle, but the sensors on the vehicle automatically collect environmental information and drive automatically based on the environmental information. The test and evaluation of autonomous driving functions is an indispensable and important link in vehicle development, technology application and commercial promotion. Vehicles with autonomous driving functions need to undergo a large number of tests to prove the stability of their various application functions and performance from the laboratory to mass production. , robustness, reliability, etc. When conducting closed site testing or actual road testing, real cars or some devices can be used to simulate traffic participants to create driving scenarios for the vehicle under test (autonomous driving vehicles), and verify the ability of the autonomous driving function of the vehicle under test to cope with the driving scenario. . However, in some specific situations, such as software and hardware failures, external environmental interference, etc., if the automatic driving function of the vehicle under test cannot be successfully controlled, the vehicle under test will collide with the real car or these devices, especially in relatively large distances. There is a greater risk of collision at high speeds, and these devices are usually not maneuverable and sensitive enough, so the test scenarios are relatively limited, for example, only lower speed tests can be performed.

Contents of the invention

This application provides vehicle testing methods and systems based on unmanned aerial vehicles. Specifically, it provides automatic driving testing methods, devices, systems, movable targets and storage media, which can provide devices with better maneuverability and sensitivity for the vehicle being tested. Create driving scenarios and test scenarios that can be richer and safer.

In a first aspect, embodiments of the present application provide an automatic driving test method. The test method includes:

Send a preset test instruction to the UAV, and the test instruction is used to instruct the UAV to move in the traffic scene, wherein the UAV carries a model of the target object, and the model of the target object follows The drone moves in the traffic scene;

Obtain the driving status of the vehicle under test in the traffic scene, wherein the vehicle under test can move autonomously in the traffic scene based on observation data of the target model and a preset automatic driving algorithm;

According to the driving state of the tested vehicle, a test result of the tested vehicle is generated.

In the second aspect, embodiments of the present application provide an automatic driving test method for a drone that can carry a model of a target object. The test method includes:

Receive test instructions;

Move in the traffic scene according to the test instruction, so that the model of the target object moves with the drone in the traffic scene.

In the third aspect, embodiments of the present application provide an automatic driving test method for the vehicle under test. The test method includes:

Based on the observation data of the target object model in the traffic scene and the preset automatic driving algorithm, the traffic scene includes a drone and a model of the target object carried by the drone. The model of the target object moves with the drone in the traffic scene;

Obtain the driving status of the vehicle under test;

The driving state of the vehicle under test is sent to the automatic driving test device, so that the automatic driving test device generates a test result of the vehicle under test according to the driving state of the vehicle under test.

In a fourth aspect, embodiments of the present application provide an automatic driving test device, including one or more processors, working individually or jointly, for executing the steps of the aforementioned automatic driving test method.

In a fifth aspect, embodiments of the present application provide a drone, including:

Flying platform for flying;

One or more processors, working individually or jointly, are used to execute the steps of the aforementioned automatic driving test method.

In a sixth aspect, embodiments of the present application provide a movable target, where the movable target includes:

The aforementioned drone;

The model of the target object can be connected to the drone and move with the drone in the traffic scene.

In a seventh aspect, embodiments of the present application provide a vehicle, including:

vehicle platform;

One or more processors, working individually or jointly, are used to execute the steps of the aforementioned automatic driving test method.

In an eighth aspect, embodiments of the present application provide an autonomous driving test system, including:

drone;

The model of the target object can be mounted on the drone and move with the drone in the traffic scene;

A vehicle under test, which can move autonomously in the traffic scene based on observation data of the model of the target object and a preset automatic driving algorithm;

The aforementioned automatic driving test device.

In a ninth aspect, embodiments of the present application provide a computer-readable storage medium. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, it causes the processor to implement the above method.

The embodiments of this application provide vehicle testing methods and systems based on unmanned aerial vehicles. Specifically, automatic driving test methods, devices, systems, movable targets and storage media are provided. UAVs, that is, unmanned aerial vehicles carry targets. The model of the target object moves with the drone in the traffic scene to simulate traffic participants, creating a test scene for the vehicle being tested. The model of the target object carried by the drone has better maneuverability and sensitivity. The response is fast and the accuracy is high. The vehicle under test driving on the ground is not easy to collide with the drone or the damage caused by the collision is light. Therefore, it can create richer test scenarios for the vehicle under test, such as higher speed testing and safer

It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and do not limit the disclosure of the embodiments of the present application.

Description of drawings

In order to explain the technical solutions of the embodiments of the present application more clearly, the drawings needed to be used in the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without exerting creative efforts.

Figure 1 is a schematic flow chart of an autonomous driving testing method provided by an embodiment of the present application;

Figure 2 is a schematic structural diagram of a model of a drone carrying a target object in one embodiment;

Figure 3 is a schematic structural diagram of a model of a UAV carrying a target object in another embodiment;

Figure 4 is a schematic structural diagram of a model of a UAV carrying a target object in yet another embodiment;

Figure 5 is a schematic diagram of the autonomous movement of the vehicle under test based on the observation of the model of the target object in one embodiment;

Figure 6 is a schematic diagram of the autonomous movement of the vehicle under test based on observation of a model of the target object in another embodiment;

Figure 7 is a schematic diagram of a drone performing an avoidance mission in an embodiment;

Figure 8 is a schematic structural diagram of a target model in an embodiment;

Figure 9 is a schematic diagram of a drone performing an avoidance mission in another embodiment;

Figure 10 is a schematic diagram of an environmental condition simulation device mounted on a drone in one embodiment;

Figure 11 is a schematic flow chart of an autonomous driving testing method provided by another embodiment of the present application;

Figure 12 is a schematic flow chart of an autonomous driving testing method provided by yet another embodiment of the present application;

Figure 13 is a schematic block diagram of an automatic driving test device provided by an embodiment of the present application;

Figure 14 is a schematic block diagram of a drone provided by an embodiment of the present application;

Figure 15 is a schematic block diagram of a movable target provided by an embodiment of the present application;

Figure 16 is a schematic block diagram of a vehicle provided by an embodiment of the present application;

Figure 17 is a schematic block diagram of an automatic driving test system provided by an embodiment of the present application.

Detailed ways

The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of this application.

The flowcharts shown in the accompanying drawings are only examples and do not necessarily include all contents and operations/steps, nor are they necessarily performed in the order described. For example, some operations/steps can also be decomposed, combined or partially merged, so the actual order of execution may change according to actual conditions.

Some embodiments of the present application will be described in detail below with reference to the accompanying drawings. The following embodiments and features in the embodiments may be combined with each other without conflict.

With the development of science and technology and the application of artificial intelligence technology, autonomous driving technology has developed rapidly and been widely used. Autonomous driving vehicles are equipped with advanced on-vehicle sensors, controllers, actuators and other devices, and integrate modern communication and network, artificial intelligence and other technologies to realize intelligent information exchange and sharing between vehicles, roads, people, cloud, etc. , equipped with functions such as complex environment perception, intelligent decision-making, and collaborative control, can achieve "safe, efficient, comfortable, energy-saving driving, and ultimately realize a new generation of cars that can replace human operation."

Based on the driving automation level of the vehicle, the existing SAE J3016 standard divides driving automation into six levels, namely L0-L5 levels, which are No Automation (L0) and Driver Assistance (L1). , Partial Automation (L2), Conditional Automation (L3), High Automation (L4) and Full Automation (L5). As the level of driving automation continues to improve, human participation in driving activities is getting lower and lower. It is foreseeable that more self-driving vehicles will be driving on the road in the future, resulting in a situation where self-driving vehicles and human-driven vehicles will run side by side on the road.

Testing and evaluation is an indispensable and important link in the development, technology application and commercial promotion of autonomous driving functions of autonomous vehicles. Different from traditional cars, the test and evaluation object of autonomous vehicles becomes the strongly coupled human-vehicle-environment-task system. As the level of driving automation increases, the functions implemented by different levels of automation increase step by step, making it extremely challenging to test and verify them. Some countries and regions have introduced corresponding laws and regulations to allow self-driving vehicles to be tested on roads. Fully verify the safety of autonomous vehicles. In addition to road testing, government agencies, scientific research institutes, and related enterprises in various countries have carried out a large amount of research work on the standard system and related evaluation methods required for the testing and evaluation of autonomous vehicles.

Autonomous driving vehicles generally undergo virtual simulation testing, closed site testing, and real road testing according to different needs. Among them, virtual simulation testing can cover all predictable scenarios within the Operational Design Domain (ODD), including corner scenes that are not easy to occur, and cover all autonomous driving functions within the ODD; closed site testing can cover all scenarios within the ODD. Extreme scenarios, such as safety-related accident scenarios and dangerous scenarios, cover the typical functions of the automatic (assisted) driving system under normal conditions and verify the simulation test results; real road tests can cover roads with typical scenario combinations within the ODD range, covering random scenarios and A combination of random elements to verify the ability of the autonomous driving function to cope with random scenarios.

Virtual simulation testing cannot simulate actual traffic scenarios very well. When the vehicle under test is tested in a closed site or on a real road, under some specific circumstances, such as software and hardware failure, external environment interference, etc., the automatic driving of the vehicle under test will fail. If the function cannot be successfully controlled, it may collide with the test vehicle. Using a real test car driven by a real person for testing is highly dangerous. If the vehicle under test cannot be successfully controlled, it may cause a car crash and death. Currently, some test vehicles are equipped with a foam panel body on a chassis with a relatively low expansion. When testing, if the vehicle under test makes a mistake in controlling the vehicle and collides with the test vehicle, the foam panel body will be scattered, and the vehicle under test can be removed from the vehicle. The chassis is rolled over. Some chassis can be passively compressed and lowered by the rolling over of the vehicle under test, allowing the vehicle under test to pass smoothly and reducing the damage and loss of the vehicle under test in dangerous scenes. However, when testing at higher speeds , there is still a high risk; and the chassis structure is complex, with contraction springs and other structures, which has the disadvantages of high cost, high price, and inconvenient operation.

To this end, the embodiments of the present application provide vehicle testing methods and systems based on unmanned aerial vehicles. Specifically, automatic driving test methods, devices, systems, movable targets and storage media are provided through unmanned aerial vehicles, that is, unmanned aerial vehicles. A model of the target object. The model of the target object moves with the drone in the traffic scene to simulate traffic participants, creating a test scenario for the vehicle being tested. The model of the target object carried by the drone has better maneuverability and sensitivity. Good, fast response and high accuracy. The vehicle under test driving on the ground is not easy to collide with the drone or the collision damage is light. Therefore, it can create richer test scenarios for the vehicle under test. For example, it can conduct higher-speed tests and more Safety.

Please refer to Figure 1. Figure 1 is a schematic flow chart of a vehicle testing method based on an unmanned aerial vehicle, that is, an autonomous driving testing method, provided in an embodiment of the present application. The automatic driving test method can be applied in an automatic driving test device to instruct the UAV to move in the traffic scene, so that the model of the target object carried by the UAV moves with the UAV in the traffic scene. , creating a test scenario for the vehicle under test, and generating test results based on the driving status of the vehicle under test in the test scenario. Among them, the tested vehicles may include vehicles with different autonomous driving levels, such as vehicles at any level from L0 to L5. It is understandable that the tested vehicles may be manned vehicles or unmanned vehicles.

The UAV may be a rotary-wing UAV, such as a four-rotor UAV, a six-rotor UAV, an eight-rotor UAV, or a fixed-wing UAV.

Among them, the automatic driving test device can be set up on the drone, of course, it can also be set up on the vehicle being tested, or it can also be set up on the roadside equipment (Road Side Unit, RSU) or terminal equipment, where the roadside equipment is a smart road The core of the system serves to connect roadside facilities and transmit road information to vehicles and the cloud. It can realize background communication functions, information broadcast functions, and high-precision positioning foundation enhancement functions; terminal devices can include mobile phones, tablets, laptops, At least one of desktop computers, remote controls, etc. In some embodiments, the terminal device can also generate corresponding test instructions according to the user's operation, and send the generated test instructions to the drone, so that the drone moves according to the test instructions, for example, the drone moves according to the test instructions. Preset test trajectory movement.

As shown in Figure 1, the automatic driving test method according to the embodiment of the present application includes steps S110 to S130.

S110. Send a preset test instruction to the UAV. The test instruction is used to instruct the UAV to move in the traffic scene. The UAV is equipped with a model of the target object, and the model of the target object is carried by the UAV. The drone moves in the traffic scene.

Please refer to FIGS. 2 to 4 . FIGS. 2 to 4 are schematic structural diagrams of a model 120 carrying a target object on the UAV 110 in different embodiments. In some embodiments, the UAV 110 and the model 120 of the target together may be referred to as the movable target 100 . Of course, the movable target 100 is not limited to including the UAV 110 and the model 120 of the target, for example A connector 130 for connecting the drone 110 to the model 120 of the target may also be included.

FIG. 5 shows a schematic diagram of autonomous movement of the vehicle under test 200 based on observation of a model of a target on a movable target 100 in one embodiment.

The drone in the movable target receives the test instruction and moves according to the test instruction. In some embodiments, the test instruction is used to instruct the drone to move in a preset test trajectory in a traffic scene. Illustratively, the test instruction is used to instruct the drone to drive in a straight line above a certain lane. The certain lane is the same or different from the lane of the vehicle under test. The certain lane is the same as the lane of the vehicle under test. The lanes of the vehicles under test may be parallel or intersecting; for example, the test instruction is used to instruct the drone to change from above one lane to above another lane; for example, the test command is used to instruct the drone to The machine moves behind the vehicle under test through the lane on the left or right side of the vehicle under test to the front of the vehicle under test, and is certainly not limited to this. For example, the test instruction can also be used to indicate the speed of the drone, for example, the speed of the drone on the test track. The speeds at different locations on the test track may be the same or different.

S120. Obtain the driving status of the vehicle under test in the traffic scene, where the vehicle under test can move autonomously in the traffic scene based on observation data of the target model and a preset automatic driving algorithm.

The model of the target in the movable target moves with the drone in the traffic scene, and the vehicle under test can move autonomously in the traffic scene based on the observation data of the model of the target and the preset automatic driving algorithm, It should be noted that the observation data of the traffic scene by the vehicle under test is not limited to the observation data of the target object model. For example, it can also include observation data of other traffic participants, roadside construction facilities, traffic signs, etc. The vehicle under test It can move autonomously based on the observation data of the traffic scene based on the preset automatic algorithm.

In some embodiments, the UAV includes an environmental sensor, and obtaining the driving status of the vehicle under test in the traffic scene includes: obtaining image information in the traffic scene through the environmental sensor of the UAV. , determining the driving state of the vehicle under test based on the image information. For example, the environmental sensors carried by the drone include radar and/or visual sensors. The radar includes at least one of the following: lidar, ultrasonic radar, millimeter wave radar, etc. The visual sensor includes at least one of the following: Binocular camera, wide-angle camera, infrared camera, etc.

Exemplarily, the UAV is equipped with the environmental sensor through a pan/tilt device, and the method further includes: controlling the pan/tilt to adjust the attitude according to the driving state of the vehicle under test, so that the vehicle under test is in a within the sensing range of the environmental sensor. Of course, it is not limited to this. For example, the visual sensor, such as the focal length and field of view of the camera, can also be adjusted according to the driving state of the vehicle under test.

In some embodiments, the vehicle under test includes a first communication module, the UAV includes a second communication module, and the UAV can communicate with the UAV through the first communication module and the second communication module. The communication connection of the vehicle under test; the obtaining the driving status of the vehicle under test in the traffic scene includes: obtaining the driving status of the vehicle under test sent by the vehicle under test through the communication connection with the vehicle under test state.

In some embodiments, as shown in Figure 5, the traffic scene is provided with a roadside device 300, the roadside device is used to obtain the driving status of the vehicle under test, and the drone includes a second communication module , the roadside equipment includes a third communication module, and the drone can communicate with the roadside equipment through the second communication module and the third communication module; the acquisition of objects in the traffic scene Measuring the driving state of the vehicle includes: obtaining the driving state of the vehicle under test obtained by the roadside device through a communication connection with the roadside device.

S130. Generate a test result of the tested vehicle according to the driving state of the tested vehicle.

In some embodiments, the test results of the vehicle under test include a data set obtained by performing preset processing on the driving state of the vehicle under test.

For example, the driving state of the vehicle under test is processed into data in a preset format, such as tables, curves, etc., to obtain the test results of the vehicle under test. Of course, it is not limited to this. For example, it can also be based on the vehicle under test. The driving status of the measured vehicle is determined to determine several evaluation indicators.

In some embodiments, generating a test result of the vehicle under test according to the driving state of the vehicle under test includes: generating a test result based on the motion state of the drone and the driving state of the vehicle under test. The test results of the vehicle under test. Based on the model movement of the drone carrying the target object, a test scene is created for the vehicle under test. It is expected that the different motion states of the drone can cause the vehicle under test to adjust its motion state accordingly. According to the drone The test results generated by the motion state of the human machine and the driving state of the vehicle under test can more accurately reflect the vehicle under test's observation of the environment and the performance of the automatic driving algorithm.

Exemplarily, the test results can be used to indicate at least one of the following: whether the vehicle under test and other traffic participants are safe, whether the actions performed by the vehicle under test are timely, and whether the driving behavior performed by the vehicle under test is accurate. . For example, the test results can be used to indicate whether the vehicle under test collides with the model of the target object in the test scenario created by the drone and the model of the target object, and whether the vehicle under test is braking. Whether the deceleration at the time exceeds the preset deceleration threshold, the vehicle under test can maintain a safe distance from the model of the target object when it goes around the model of the target object, etc., of course, it is not limited to this.

In some embodiments, the performance of the vehicle under test can be tested by observing a model of how the vehicle under test responds to a target object, such as an assisted driving function or an autonomous driving function. A variety of real-life scenarios can be simulated through models of drones carrying targets, which may lead to dangerous events. For example, in the Autonomous Emergency Braking (AEB) test, the target model can simulate a pedestrian suddenly jumping out from the side of the road that is not a crosswalk (or jumping out from behind a stopped truck on the road). According to the measured The response parameters of the vehicle under test, such as the response time of the vehicle under test, critical collision time, minimum braking distance, degree of collision damage, etc., are used to evaluate the performance of the AEB system; for example, in the autonomous driving test where the highway vehicle cuts into the scene, the target object The model can simulate a motor vehicle, and the critical safety boundary of the scene can be extracted by paying attention to the time to collision (TTC) of the vehicle under test at the time of entry, such as the time to collision (TTC) with the model of the target object.

Optionally, the model of the target object includes one or more of the following: a flexible plate, a flexible film, and an inflatable airbag. Exemplarily, multiple flexible boards are combined to form the front side panel, left side panel, back side panel, and right side panel of the model of the target object; Exemplarily, the model of the target object includes a frame and a flexible film, The flexible membrane is supported by the frame to form a three-dimensional model of the target object; for example, the air bag is inflated to form a three-dimensional model of the target object; for example, part of the model of the target object is a frame composed of flexible plates, and the other parts are frames Supported flexible membrane. Of course it is not limited to this. Optionally, the side walls of the flexible board, flexible film, and airbag can be metal or non-metal, and can be a single-layer structure or a multi-layer structure.

Optionally, the maximum weight of the target object model is positively related to the area covered by the paddle disc of the UAV, so that the target object model can move with the UAV in the traffic scene, and has Higher maneuverability, such as acceleration. Optionally, there may be one UAV carrying the model of the target object, or there may be multiple UAVs.

Optionally, the weight of the target model is less than or equal to a preset value. For example, the preset value ranges from 1 to 20 kilograms. For example, the weight of the target model is 1 kilogram, 2 kilograms, or 5 kilograms. Optionally, the model of the target object is made of lightweight materials, such as foam board, cardboard, porous board, etc.

Optionally, the airbag can be inflated or deflated when mounted on the drone. When the airbag is deflated, it is smaller in size, making it easier to store and transport. When inflated, the target model formed is lighter in weight, making it easier to move with the drone.

Optionally, the airbag is inflated by one or more of the following methods: inflated by the airflow generated by the blades when the drone is flying, inflated by an inflating device on the airbag, or inflated by an external device connected to the airbag. The device is inflated. The external inflatable device to which the airbag is connected may or may not be provided on the drone. When inflated in a variety of ways, the air bag's air volume can be better ensured so that the air bag can better serve traffic participants.

Optionally, the target object model can be a traffic participant, for example, different types of vehicles, people of different sizes, different animals, bicycles, motorcycles or balance vehicles, etc.

The appearance shape of the model of the target object is similar to that of traffic participants. The traffic participants include one or more of the following: motor vehicles, non-motor vehicles, pedestrians, and animals. Of course, it is not limited to this. For example, it can be on a lane. plastic bags and other debris. The external shape of the target model 120 shown in Figure 2 is similar to that of a motor vehicle. The external shape of the target model 120 shown in Figure 3 is similar to that of a non-motor vehicle. The external shape of the target model 120 is shown in Figure 4 Similar to pedestrians, the appearance characteristics of traffic participants can be simulated through the model 120 of the target object, and the movement of the traffic participants is simulated by the drone 110 carrying the model 120 of the target object. Models 120 of objects for different traffic participants can be used in different test scenarios.

Please refer to Figure 6. There are multiple movable targets 100 in the traffic scene, which are used to simulate pedestrians, motor vehicles and non-motor vehicles, and can test the vehicle 200 under test in complex test scenarios. In other embodiments, there are multiple movable targets 100 in the traffic scene for simulating motor vehicles. Some of the movable targets 100 are in the same lane as the vehicle 200 being tested, and the lanes of the remaining movable targets 100 are in the same lane as the vehicle being tested. The lanes of the test vehicle 200 are adjacent or intersecting, which can be used for the vehicle to be tested to conduct single scene or continuous scene testing in an environment with multiple traffic participants, so as to realize the coordinated work of multiple movable targets 100, such as multiple movable targets 100. The target object 100 can preset a path based on the test scenario and execute it accurately, or it can adjust the movement of the drone in real time based on the driving status of the vehicle under test (a vehicle equipped with ADAS/AD to be tested). The movable target object 100 can also Real-time obstacle avoidance based on wireless communication or self-awareness.

Optionally, the model of the target object and/or the material of all or part of the outer surface of the drone are set according to the radar type of the vehicle under test. In this way, the radar detection characteristics of the target model are similar to those of the corresponding traffic participants, and the radar detection data of the target model by the vehicle under test can be closer to the radar detection data of real traffic participants. Optionally, the detection signal of the radar is any of the following: laser, millimeter wave, ultrasonic wave, and of course is not limited to this.

Optionally, the outer surface is configured to have one or more of the following characteristics for the radar detection signal: diffuse reflection characteristics, refraction characteristics, and absorption characteristics. For example, the outer surface of the drone can absorb the detection signal of the radar, the model of the target object can reflect the detection signal of the radar, and the radar of the vehicle being tested or other traffic participants can better detect the model of the target object, It can also prevent the interference of the radar by the reflected detection signal of the drone, and the test scene is more in line with the real movement scene of the vehicle under test.

Optionally, the UAV and the model of the target object may be rotatably connected. In some embodiments, when the vehicle under test collides with the model of the target object, the model of the target object rotates relative to the UAV under the influence of the impact, which can reduce or avoid damage to the UAV caused by the impact, and Reduce or avoid collision damage to the vehicle under test. In some embodiments, when the driving state of the vehicle under test does not meet the preset driving state conditions, for example, when it may collide with the model of the target object, the drone can be controlled to adjust the position of the model of the target object. For example, the target model can be rotated relative to the UAV. The bottom end of the target model can be away from the ground and the vehicle being tested. For example, the vehicle being tested can pass under the target model to reduce or avoid impact on the target. damage caused by models and drones, as well as reduce or avoid collision damage to the vehicle under test.

Optionally, please refer to FIG. 7 , the drone 110 and the target model 120 are detachably connected. It is convenient to connect the drone 110 to the model 120 of another target object, or to connect the model 120 of the target object to another drone 110, and to facilitate storage and transportation. For example, when the model 120 of the target object is damaged or when different test scenarios need to be created, the model 120 of the target object carried by the drone 110 can be easily replaced.

Optionally, the connection between all or part of the target model and the drone can be automatically disconnected in a preset state.

In some embodiments, please refer to FIG. 7 , the connection between the target object model 120 and the drone 110 can be automatically disconnected in a preset state based on the target object model 120 and the drone 110 . This is achieved by disconnecting at least two components of the connection piece 130 between the human machine 110 and the machine 110 . Exemplarily, the connecting member 130 connects the target model 120 and the drone 110 in at least one of the following ways: snap fit, interference fit, magnetic attraction, adhesive connection, and of course Not limited to this.

In some embodiments, all or part of the connection between the model of the target object and the drone is easily detached under tension, for example, when the vehicle under test collides with the model of the target object. The pulling force disconnects all or part of the target model from the drone, reducing or avoiding damage to the target model and the drone caused by the impact, and also facilitating the drone to move away quickly. The vehicle under test, and reduce or avoid collision damage to the vehicle under test.

Optionally, the preset state includes: the pulling force between the model of the target object and the drone is greater than a preset threshold; and/or the tension between the model of the target object and the drone The pulling force direction is in the preset direction. For example, the connection structure between the model of the target object and the UAV can be designed to ensure reliable connection between the model of the target object and the UAV, and also to enable the vehicle under test to be connected to the UAV. The pulling force generated when the model of the target object collides causes all or part of the model of the target object to be disconnected from the drone. For example, when the component of the pulling force in the horizontal direction between the model of the target object and the drone is greater than a specific value, the connection between the model of the target object and the drone is disconnected.

In some embodiments, the model of the target object is connected to the drone based on a connector, and the connector disconnects the model of the target object from the drone in response to a triggering instruction to enter the preset state. The connection of the drone. Exemplarily, the drone is provided with an electromagnetic lock, which can be engaged with the corresponding structure on the model of the target object, and the electromagnetic lock can be controlled to unlock the card according to the triggering command. connection, so that the model of the target object is disconnected from the UAV.

Optionally, the triggering instruction is generated by a sensor sensing the pulling force between the model of the target object and the drone, and the sensor is provided on the model of the target object and/or the drone. The sensor includes, for example, a Hall sensor and/or a tension sensor, but is of course not limited thereto. For example, one end of the tension sensor is connected to the drone, and the other end is connected to the model of the target object. Exemplarily, part of the model of the target object is close to the preset position of the UAV under the action of pulling force. The sensor at the preset position senses the approach of the model of the target object, and the said model can be generated. Trigger command.

Optionally, as shown in FIGS. 8 and 9 , the model 120 of the target object includes a plurality of components 121 , and the plurality of components 121 are detachably connected. This facilitates assembly, storage, and transportation of the target model 120 . Exemplarily, at least two adjacent components 121 among the plurality of components 121 are connected through one or more of the following methods: snap-fit, interference fit, magnetic attraction, and adhesive connection. For example, as shown in FIGS. 8 and 9 , adjacent components 121 are detachably connected through connectors 122 .

Optionally, as shown in FIG. 9 , at least two components 121 among the plurality of components 121 are connected to different drones 110 . Multiple drones 110 can provide the target model 120 with stronger maneuverability, higher speed and acceleration.

Optionally, as shown in Figure 10, the UAV 110 can be equipped with an environmental condition simulation device 140. The environmental condition simulation device 140 can simulate one or more of the following environments: rainfall, dense fog, dust, and light.

In some embodiments, the method further includes: controlling the operation of an environmental condition simulation device mounted on the drone, so that the environmental condition simulation device simulates one or more of the following environments: rainfall, dense fog, Dust, light. As shown in Figure 10, the drone 100 can receive the control instructions sent by the terminal device 400 through the communication device 111, and control the environmental condition simulation device 140 to simulate one or more of the following environments according to the control instructions: rainfall, dense fog , dust, light. Thus, the performance of the tested vehicle under various environmental conditions can be achieved.

In some embodiments, the method further includes: when the driving state of the vehicle under test does not meet the preset driving state conditions, controlling the drone to perform a preset avoidance task so that the target object The model avoids the vehicle under test. Prevent the vehicle under test from colliding with the model of the target object, and prevent damage to the model of the target object from affecting the test progress. For example, the time spent on replacing, repairing, and assembling the model of the target object can be reduced or avoided. Reduce the cost of autonomous driving testing and enable more accurate testing. Not only can testing be performed in a single scenario, but in some implementations, autonomous driving functions can also be tested in continuous operating scenarios to support high-level autonomous driving system testing. verify.

In some embodiments, controlling the UAV to perform a preset avoidance task so that the model of the target object avoids the vehicle under test includes: controlling the UAV in a horizontal direction and/or The motion state is adjusted in the vertical direction so that the model of the target object avoids the vehicle under test.

Exemplarily, the model of the target object moves in front of the lane where the vehicle under test is located, and when the vehicle under test is about to collide with the model of the target object, the drone is controlled to accelerate forward so that the vehicle under test moves forward. The model of the target object is far away from the vehicle under test.

Exemplarily, as shown in FIG. 7 , the drone 110 moves upward in the vertical direction, taking the model 120 of the target object off the ground, so that the vehicle under test moves away from the model 120 of the target object. Pass underneath, or reduce the damage caused during a collision.

For example, the acceleration of the drone when adjusting its motion state in the vertical direction is determined based on the weight of the model of the target object. For example, when the weight of the model of the target object is heavier, the acceleration is lower so that the drone can successfully carry the model of the target object away from the vehicle under test.

In some embodiments, controlling the UAV to perform a preset avoidance task so that the model of the target object avoids the vehicle under test includes: controlling the UAV to adjust the position of the target object. The pose of the model is such that the bottom end of the model of the target object is away from the ground. For example, by rotating the model of the target relative to the drone, the bottom end of the model of the target can be away from the ground and the vehicle being tested. For example, the vehicle being tested can pass under the model of the target, reducing or avoiding impact on the target. Damage caused by models and drones, as well as reducing or avoiding collision damage to the vehicle under test.

In some implementations, please refer to FIG. 8 , the controlling the UAV to perform a preset avoidance task so that the model of the target object avoids the vehicle under test includes: controlling the UAV to perform The preset avoidance task is to separate the multiple detachably connected components of the model of the target object from each other. The separation of multiple components can reduce losses in the event of a collision with the vehicle being tested.

For example, after the multiple components are separated from each other, the number of components located on the driving path of the vehicle under test is reduced. The vehicle under test collides with fewer components when traveling along the driving path, or passes through gaps after components are separated, thereby reducing or avoiding damage caused by collisions.

Exemplarily, the plurality of components are separated from each other under the action of the downward pressure wind field of the UAV blades when the UAV performs the avoidance mission. For example, the magnitude and/or direction of the power output of the UAV can be adjusted so that the downward pressure wind field of the UAV's blades faces the model of the target object, so that the target object The multiple detachably connected parts of the model are separated from each other under the action of the downward pressure wind field.

Exemplarily, the plurality of components are separated from each other under the driving action of the mechanical structure when the UAV performs the avoidance mission. For example, the mechanical structure includes a driving device and a plurality of support arms connected to the driving device. At least two of the plurality of components are connected to different support arms. The driving device can drive the support. The arm moves to separate the at least two components from each other. The drive may include, for example, an electric drive and/or an elastic drive.

For example, the plurality of components can be used to control the effects of the UAV blades on the wind field when the UAV performs the avoidance mission and the mechanical structure of the UAV when the UAV performs the avoidance mission. separated from each other by driving. The longer the separation distance, the wider the gap left for the vehicle to be tested to pass.

In some embodiments, please refer to FIG. 9 , the control of the drone to perform a preset avoidance mission to separate multiple detachably connected components of the target model includes: controlling A plurality of drones move away from the vehicle under test in different directions, so that the multiple components of the model of the target object carried by the drones are separated from each other and move away from the vehicle under test in different directions. . As shown in Figure 9, the drone on the left moves to the left with the components on the left, and the drone on the right moves to the right with the components on the right, away from the vehicle being tested, leaving the vehicle to be tested. Measure the clearance for vehicle traffic.

In some embodiments, the connection between adjacent components is easy to detach under the action of tension, such as under the action of a wind field or the driving action of a mechanical structure, or when different drones move in different directions. The connection between adjacent components is disconnected under the action of tension, so that the components can move away from the vehicle under test.

In other embodiments, connections between adjacent components can be broken based on control instructions. For example, when controlling the drone to perform a preset avoidance mission, the connector between adjacent components can be controlled to disconnect the adjacent components. After the connection between adjacent components is disconnected, it can move away from the vehicle under test faster under the action of the wind field or the driving action of the mechanical structure, or the pulling force when different drones move in different directions.

Optionally, the method further includes: when the driving state of the vehicle under test does not meet the preset driving state conditions, disconnecting all or part of the target model from the drone. Exemplarily, when the driving state of the vehicle under test does not meet the preset driving state conditions, the connection member used to connect the target object to the drone is triggered to disconnect the target object model. Said connection to said drone. It can prevent the target model from affecting the safety of the drone when it collides with the vehicle being tested, and it can also reduce the damage caused by the collision to the vehicle being tested. The model of the target object is low in cost or easy to assemble, so it can reduce the damage and time cost caused by collision.

In some embodiments, whether the driving state of the vehicle under test satisfies the driving state condition is determined based on the driving state of the vehicle under test obtained in one sampling period, or based on the driving state obtained in multiple sampling periods. It is determined by the changing trend of the driving state of the vehicle under test. The one sampling period may be the latest sampling period, or one of the plurality of sampling periods. For example, the driving state with the highest confidence among the driving states sampled in the multiple sampling periods is determined, and the confidence level is determined. The driving state with the highest degree of accuracy is the driving state of the vehicle under test. For example, based on the changing trend of the driving state of the vehicle under test obtained in multiple sampling periods, it can be determined whether the automatic driving function of the vehicle under test is invalid or unreliable.

In some embodiments, the driving state of the vehicle under test includes at least one of the following: motion parameters of the vehicle under test, observation information of the traffic scene by the vehicle under test, control of the vehicle under test Information and the relative motion relationship between the vehicle under test and other objects in the traffic scene are of course not limited to this. When one or more driving states of the vehicle under test do not meet the preset driving state conditions, the drone is controlled to adjust the driving state so that the model of the target object moves away from the vehicle under test.

Exemplarily, obtaining the driving status of the vehicle under test in the traffic scene includes: obtaining observation information of the traffic scene collected by an environmental sensor mounted on the vehicle under test, and/or obtaining the vehicle under test. The control information of the vehicle under test is determined by the vehicle's automatic driving module.

For example, the motion parameters of the vehicle under test include at least one of the following: speed, acceleration, and position. The position is, for example, the lane in which it is located.

For example, the vehicle under test can obtain observation information of the traffic scene through one or more of environmental sensors, such as cameras, millimeter-wave radar, and lidar. The observation information includes at least one of the following: Target Tracking information, lane line identification information, drivable area information, traffic flow information.

For example, the vehicle under test can determine the control information of the vehicle under test through the automatic driving module. The control information includes, for example, at least one of the following: trajectory planning information, behavior interpretation information, diagnostic information, braking signal, Turn signals, acceleration signals, human-computer interaction warning information.

For example, the relative motion relationship between the vehicle under test and other objects in the traffic scene includes at least one of the following: relative position relationship, relative speed, relative acceleration, and of course is not limited to this. It should be noted that the relative motion relationship can be obtained by the vehicle under test, or by the drone, or by the roadside equipment. Of course, it can also be obtained based on the vehicle under test. The driving state and the flight state of the UAV are determined, for example, the relative motion relationship between the vehicle under test and the UAV is determined based on the position of the vehicle under test and the position of the UAV.

In some embodiments, whether the driving state of the vehicle under test satisfies the driving state condition is determined based on the relative positional relationship between the vehicle under test and the model of the target object. The relative positional relationship between the vehicle under test and the model of the target object can be determined based on the relative positional relationship between the vehicle under test and the drone. For example, the relationship between the vehicle under test and the drone can be determined. The relative positional relationship is determined as the relative positional relationship between the vehicle under test and the model of the target object.

For example, when the relative distance between the vehicle under test and the model of the target object is less than or equal to a preset value, such as 1 meter, it is determined that the driving state of the vehicle under test does not meet the preset driving state. According to the conditions, the drone can be controlled to perform a preset avoidance task, so that the model of the target object avoids the vehicle under test. When the distance between the vehicle being tested and the model of the target object is less than a safe distance, the drone can be controlled in time to perform a preset avoidance task, so that the model of the target object is far away from the vehicle being tested to prevent collision. .

For example, when the relative distance between the vehicle under test and the model of the target object is less than or equal to a preset value, and the vehicle under test and the model of the target object have moved or are about to move in parallel, If it is determined that the driving state of the vehicle under test does not meet the preset driving state conditions, the drone can be controlled to perform a preset avoidance task so that the model of the target object avoids the vehicle under test. Collisions can be prevented more accurately.

In other embodiments, whether the driving state of the vehicle under test satisfies the driving state condition is determined based on a preset type of evaluation index, and the evaluation index is determined based on the driving state of the vehicle under test. . For example, the value of a preset evaluation index can be determined according to one or more driving states of the vehicle under test, and whether the driving state satisfies the driving state condition can be determined based on the value of the evaluation index. For example, the evaluation index When the value exceeds the preset range, it is determined that the driving state of the vehicle under test does not meet the preset driving state conditions, and the UAV can be controlled to perform a preset avoidance task so that the model of the target object avoids the target. Describe the vehicle under test.

For example, the evaluation index is determined based on the relative position relationship and relative speed of the vehicle under test and the model of the target object. For example, the preset type of evaluation index includes: collision time and/or collision critical deceleration determined based on the relative position relationship and relative speed of the vehicle under test and the model of the target object.

Wherein, the time to collision (TTC) can be determined based on the following time: when the model of the vehicle under test and the target object maintains the relative speed, the time from the relative position relationship to the occurrence of collision , for example, the collision time may be determined based on the ratio of the relative distance to the relative speed of the vehicle under test and the model of the target object. At the current moment, the speed of the vehicle behind is greater than that of the vehicle in front. If the two vehicles keep their original speed and driving trajectory unchanged (that is, assuming that the driver or the autonomous driving system does not take avoidance actions), based on the current speed and trajectory, they will be at a certain point. A collision occurs at a moment, then the time period from the current moment to the collision is the collision time. The smaller the collision time, the greater the likelihood of an accident. For example, when the collision time is less than or equal to a preset time threshold, it is determined that the driving state of the vehicle under test does not meet the preset driving state conditions.

Wherein, the critical collision deceleration rate (Deceleration Rate to Avoid a Crash, DRAC) is determined according to the following deceleration rate: when the model of the tested vehicle and the target object just avoids a collision, the deceleration rate required by the tested vehicle is The speed or the deceleration required by the model of the target object. For example, the critical collision deceleration is determined based on the ratio of the square of the relative speed and the relative distance between the vehicle under test and the model of the target object. For two vehicles that are closely following each other, if the speed of the following vehicle is greater than that of the leading vehicle, the deceleration required by the following vehicle to avoid rear-end collision with the leading vehicle is the collision critical deceleration, or it can be called the deceleration to avoid a rear-end collision. Speed. When the critical collision deceleration of the rear vehicle exceeds the vehicle performance or the maximum available decelerate (MADR) that the passengers can withstand, there is a high probability of a collision. For example, when the collision critical deceleration is greater than or equal to a preset deceleration threshold, it is determined that the driving state of the vehicle under test does not meet the preset driving state conditions.

Exemplarily, the preset type of evaluation index includes: when the vehicle under test is in the driving state, the probability of collision between the vehicle under test and the model of the target object. For example, the probability of collision can be determined based on the driving state of the vehicle under test, or based on the driving state of the vehicle under test and the flight state of the drone. For example, the probability of collision can be determined based on a machine learning target object model. Probability of collision. For example, if the probability of the vehicle under test colliding with the model of the target object when the vehicle under test maintains the driving state is greater than or equal to the probability threshold, it is determined that the driving state of the vehicle under test does not satisfy the driving conditions.

In some other embodiments, whether the driving state of the vehicle under test satisfies the driving state condition is determined based on the obtained driving state of the vehicle under test and the expected driving state of the vehicle under test, so The expected driving state is determined based on the test instructions.

Exemplarily, the drone moves in the traffic scene according to the test instruction, and creates a test scene for the vehicle under test. If the vehicle under test can drive in the expected driving state corresponding to the test instruction, Then at least it will not collide with the model of the target object. It can be understood that the expected driving state is a driving state in which the vehicle under test can avoid collision with the model of the target object when the drone moves in the traffic scene according to the test instructions.

In some embodiments, the expected driving state may be determined according to a user's setting operation, for example. Of course, it is not limited to this. For example, it can be determined through statistical analysis of a large number of traffic scene images.

For example, if the driving state of the vehicle under test obtained in step S120 does not match the expected driving state of the vehicle under test, it may be determined that the driving state of the vehicle under test does not meet the preset driving state conditions, If there is a risk of collision, the drone can be controlled to perform a preset avoidance task, so that the model of the target object avoids the vehicle under test.

For example, if the drone decelerates in front of the lane where the vehicle under test is located according to the test instruction, the expected driving state of the vehicle under test includes negative longitudinal acceleration and/or non-zero lateral acceleration. When the drone decelerates directly in front of the vehicle under test, if the vehicle under test can slow down and/or change lanes, it can avoid collision with the model of the target object; however, if the vehicle under test does not actually decelerate and change lanes, it will If the vehicle speed remains unchanged or accelerates, there is a risk of collision, and it is determined that the driving state of the vehicle under test does not meet the preset driving state conditions, and the drone can be controlled to perform a preset avoidance task so that all The model of the target object avoids the vehicle under test. Optionally, the expected acceleration range can also be determined according to the test instructions. If the actual acceleration of the vehicle under test exceeds the expected acceleration range, it can also be determined that the driving state of the vehicle under test does not meet the preset driving conditions. status conditions.

For example, if the drone decelerates in front of the vehicle under test according to the test instruction, the expected driving state of the vehicle under test includes: the prompt module mounted on the vehicle under test outputs the first prompt information . When the drone decelerates directly in front of the vehicle under test, if the prompt module of the vehicle under test outputs the first prompt information, the driver can be reminded to control the vehicle under test to slow down and/or change lanes to avoid collision with the model of the target object. . If the prompt module of the vehicle under test does not output the first prompt information, there is a risk of collision, and it is determined that the driving state of the vehicle under test does not meet the preset driving state conditions. It can be understood that in some embodiments, the driving status of the vehicle under test may also include prompt information output by the vehicle under test.

For example, if the drone drives into the current lane of the vehicle under test at the left front or right front of the vehicle under test according to the test instruction, the expected driving state of the vehicle under test includes: Negative longitudinal acceleration and/or non-zero lateral acceleration. When the drone merges with the vehicle under test at a position close to the vehicle under test, if the vehicle under test can slow down and/or change lanes, it can avoid collision with the model of the target object; However, if the vehicle under test does not actually decelerate or change lanes, but maintains the same speed or accelerates, there is a risk of collision. It is determined that the driving state of the vehicle under test does not meet the preset driving state conditions, and all the driving conditions can be controlled. The UAV performs a preset avoidance task so that the model of the target object avoids the vehicle under test. Optionally, the expected acceleration range can also be determined according to the test instructions. If the actual acceleration of the vehicle under test exceeds the expected acceleration range, it can also be determined that the driving state of the vehicle under test does not meet the preset driving conditions. status conditions.

In some embodiments, the subsequent action trend of the vehicle under test can be determined based on the vehicle under test's observation information of the traffic scene, and the flight status of the drone corresponding to the test instruction and the vehicle The subsequent action trend determines whether there is a risk of collision and whether the driving state of the vehicle under test meets the preset driving state conditions.

In some embodiments, the expected value of the vehicle under test's observation information of the traffic scene can be determined according to the test instruction. When the actual observation value of the vehicle under test is the same as or close to the expected value, Collision with the model of the target object can be avoided. If the actual observed value of the vehicle under test is different from the expected value, there is a probability of collision, and it can be determined that the driving state of the vehicle under test does not meet the preset driving conditions. status conditions.

For example, if the drone decelerates in front of the lane where the vehicle under test is located according to the test instruction, the expected driving state of the vehicle under test includes at least one of the following: The acceleration of the model of the target object is negative, and the relative distance between the model of the target object and the vehicle under test observed by the vehicle under test decreases.

In some embodiments, the subsequent action trend of the vehicle under test can be determined based on the control information output by the automatic driving module of the vehicle under test, and based on the driving state of the model of the target object corresponding to the test instruction and the The subsequent movement trend of the vehicle determines whether there is a risk of collision, and whether the driving state of the vehicle under test meets the preset driving state conditions.

In some embodiments, the expected value of the control information of the vehicle under test can be determined according to the test instruction. When the actual control information of the vehicle under test is the same as or close to the expected value, the interaction with the control information of the vehicle under test can be avoided. When the model of the target object collides, if the actual control information of the vehicle under test is different from the expected value, there is a probability of collision, and it can be determined that the driving state of the vehicle under test does not meet the preset driving state conditions.

For example, if the drone decelerates in front of the lane where the vehicle under test is located according to the test instruction, the expected driving state of the vehicle under test includes: controlling the vehicle under test to decelerate and/or Lane changing control information.

Optionally, the driving state of the vehicle under test includes the state of the execution system of the vehicle under test; the execution system of the vehicle under test includes at least one of the following: a braking system, a steering system, and a driving system. The driving trajectory of the tested vehicle can be determined according to the state of the execution system of the tested vehicle, and whether it will collide with the model of the target object is determined according to the driving trajectory. If it is determined that it will collide with the model of the target object, , it can be determined that the driving state of the vehicle under test does not meet the preset driving state conditions.

For example, if the drone decelerates in front of the lane where the vehicle under test is located according to the test instructions, the expected driving state of the vehicle under test includes at least one of the following: braking by the braking system; The steering system turns and the drive system reduces the accelerator.

For example, if the drone drives into the current lane of the vehicle under test at the left or right front of the vehicle under test according to the test instruction, the expected driving state of the vehicle under test includes at least the following: One: the braking system brakes, the steering system turns, and the driving system reduces the accelerator.

For example, when the drone decelerates directly in front of the vehicle under test or drives into the current lane of the vehicle under test in front of the left or right front of the vehicle under test, if the vehicle under test actually executes the state of the system According to the expected value of the state of the execution system corresponding to the test instruction, the vehicle under test can be slowed down or reduced in acceleration or changed lanes, thereby avoiding a collision with the model of the target object; but if the actual execution system of the vehicle under test is The state of the vehicle is different from the expected value of the state. If the vehicle under test does not actually decelerate or change lanes, but maintains the same speed or accelerates, there is a risk of collision, so it can be determined that the driving state of the vehicle under test does not meet the expected value. According to the set driving state conditions, the drone can be controlled to perform a preset avoidance task, so that the model of the target object avoids the vehicle under test.

In some other embodiments, whether the driving state of the vehicle under test satisfies the driving state condition is based on the observation information of the traffic scene by the vehicle under test and the observation of the traffic scene by the drone. Whether the observation information is consistent is determined. For example, if the observation information of the traffic scene by the vehicle under test is inconsistent with the observation information of the traffic scene by the drone, there is a risk of collision, so that the vehicle under test can be determined. If the driving state does not meet the preset driving state conditions, the drone can be controlled to perform a preset avoidance task so that the model of the target object avoids the vehicle being tested; if the vehicle being tested is dangerous to the traffic If the observation information of the scene is consistent with the observation information of the traffic scene by the drone, it can be determined that the observation information of the vehicle being tested is also accurate, accurate decisions and control can be made based on the observation information, and the possibility of collision is relatively high. Low. For example, if the drone is driving in front of the lane where the vehicle under test is located in accordance with the test instructions, it observes a traffic light showing a red light or a speed limit sign not far ahead, but the vehicle under test does not observe it. At a red light or speed limit sign, if the drone and the target model decelerate and brake but the vehicle under test does not decelerate and brake, there is a risk of collision, so it can be determined that the driving state of the vehicle under test does not meet the preset requirements. driving conditions.

The automatic driving test method provided by the embodiment of the present application uses a drone to carry a model of a target object. The model of the target object moves with the drone in the traffic scene to simulate traffic participants, creating a test scene for the vehicle under test. , the model of the target carried by the drone has better maneuverability and sensitivity, fast response, and high accuracy. The vehicle under test driving on the ground is not easy to collide with the drone or the collision loss is light, so it can create a model for the vehicle under test. Richer test scenarios, such as higher-speed testing, and safer testing.

The autonomous driving test method provided by the embodiments of this application is designed through various escape methods to avoid collision or almost no loss after the collision. It can conduct continuous simulation tests of multiple groups of dangerous scenes and has the characteristics of high efficiency, continuous scene simulation, and safety. Etc.

Optionally, the model of the target object is made of a small amount of inflated soft plates or films, which has the advantages of low cost and easy splicing.

Optionally, the vehicle under test can communicate with the movable target in real time, so that the movable target can more accurately simulate the real scene according to the driving status of the vehicle under test.

Optionally, for complex traffic flow test scenarios, there is no need to combine other physical vehicles and expensive auxiliary driving equipment to complete the multi-vehicle collaboration scenario safely and efficiently.

Optionally, complex test scenarios involving pedestrians, motor vehicles, and non-motor vehicles can be completed. Pedestrians, motor vehicles, and non-motor vehicles can communicate and sense in real time, simulating real complex street conditions.

Optional, not only can key parameters and motion paths be set in advance, but it can also meet the perceptual communication function during testing, self-adjustment for real-time changes, collaborative changes of multiple vehicles, mutual reference, and mutual adjustment.

Optionally, it can simulate complex environmental conditions, such as rainfall, dense fog, dust, light, etc., and can verify the expected functional safety in harsh environments during testing.

Optional, the test system based on the drone has relatively high mobility sensitivity. The response maneuverability of the drone is better than that of the ground mobile platform, and the maximum deceleration and speed are sufficient (extreme dangerous working conditions can be simulated).

Please refer to FIG. 11 in conjunction with the above embodiment. FIG. 11 is a schematic flow chart of a vehicle testing method based on an unmanned aerial vehicle, that is, an autonomous driving testing method, provided in another embodiment of the present application.

The automatic driving test method is used for a drone capable of carrying a model of a target object. The testing method includes steps S210 to S220.

S210. Receive test instructions.

S220: Move in the traffic scene according to the test instruction, so that the model of the target object moves with the drone in the traffic scene.

The specific principles and implementation methods of the automatic driving test method provided by the embodiments of this application are similar to the automatic driving test method of the previous embodiments, and will not be described again here.

Please refer to FIG. 12 in conjunction with the above embodiment. FIG. 12 is a schematic flow chart of a vehicle testing method based on an unmanned aerial vehicle, that is, an autonomous driving testing method, provided in yet another embodiment of the present application.

The automatic driving test method is used for the vehicle under test, and the test method includes steps S310 to S330.

S310. Move autonomously in the traffic scene based on the observation data of the model of the target object in the traffic scene and the preset automatic driving algorithm. The traffic scene includes a drone and a model of the target object carried by the drone, The model of the target object moves with the drone in the traffic scene;

S320. Obtain the driving status of the vehicle under test;

S330. Send the driving state of the vehicle under test to the automatic driving test device, so that the automatic driving test device generates test results of the vehicle under test according to the driving state of the vehicle under test.

The specific principles and implementation methods of the automatic driving test method provided by the embodiments of this application are similar to the automatic driving test method of the previous embodiments, and will not be described again here.

Please refer to FIG. 13 in conjunction with the above embodiment. FIG. 13 is a schematic block diagram of the automatic driving test device 500 provided by the embodiment of the present application.

Among them, the automatic driving test device 500 can be installed on a drone, of course, it can also be installed on a vehicle under test, or it can also be installed on a roadside unit (Road Side Unit, RSU) or terminal equipment. When the automatic driving test device is installed on a drone, it can control the drone to adjust its trajectory faster, more real-time and more accurately.

The automatic driving test device 500 includes one or more processors 501 , and the one or more processors 501 work individually or jointly to execute the aforementioned automatic driving test method.

Exemplarily, the automatic driving test device 500 further includes a memory 502 .

For example, the processor 501 and the memory 502 are connected through a bus 503, such as an I2C (Inter-integrated Circuit) bus.

Specifically, the processor 501 may be a micro-controller unit (MCU), a central processing unit (Central Processing Unit, CPU) or a digital signal processor (Digital Signal Processor, DSP), etc.

Specifically, the memory 502 may be a Flash chip, a read-only memory (ROM, Read-Only Memory) disk, an optical disk, a USB disk, a mobile hard disk, or the like.

The processor 501 is used to run a computer program stored in the memory 502, and implement the aforementioned automatic driving test method when executing the computer program.

The specific principles and implementation methods of the automatic driving test device 500 provided by the embodiment of the present application are similar to the automatic driving test method of the previous embodiment, and will not be described again here.

Please refer to Figure 14 in conjunction with the above embodiment. Figure 14 is a schematic block diagram of an unmanned aerial vehicle provided by an embodiment of the present application, that is, a UAV 600.

The drone 600 includes a flying platform 610, which is used for flying. The UAV 600 also includes one or more processors 601, and the one or more processors 601 work individually or jointly to perform the steps of the aforementioned automatic driving test method.

The specific principles and implementation methods of the UAV 600 provided by the embodiments of this application are similar to the automatic driving test method of the previous embodiments, and will not be described again here.

Please refer to FIG. 15 in conjunction with the above embodiment. FIG. 15 is a schematic block diagram of a movable target 700 provided by an embodiment of the present application.

The movable target 700 includes the aforementioned drone 600 and a model 710 of the target. The model 710 of the target can be connected to the drone and move with the drone in the traffic scene.

The specific principles and implementation methods of the movable target 700 provided by the embodiment of the present application are similar to the UAV 600 of the previous embodiment, and will not be described again here.

Please refer to FIG. 16 in conjunction with the above embodiment. FIG. 16 is a schematic block diagram of the vehicle 800 provided by the embodiment of the present application.

The vehicle 800 includes a vehicle platform 810, and one or more processors 801. The one or more processors 801 work individually or jointly to perform the steps of the aforementioned autonomous driving test method.

The specific principles and implementation methods of the vehicle 800 provided by the embodiment of the present application are similar to the automatic driving test method of the previous embodiment, and will not be described again here.

Please refer to FIG. 17 in conjunction with the above embodiment. FIG. 17 is a schematic block diagram of a vehicle testing system based on an unmanned aerial vehicle, or an autonomous driving testing system 900, provided by an embodiment of the present application.

Autonomous driving test system 900, including:

Drone 910;

The model 920 of the target object can be mounted on the drone 910 and move with the drone 910 in the traffic scene;

The vehicle under test 930 can move autonomously in the traffic scene based on the observation data of the target model and the preset automatic driving algorithm;

The aforementioned automatic driving test device 500.

In some implementations, the autonomous driving test device 500 may be disposed on the drone 910 .

The specific principles and implementation methods of the autonomous driving test system 900 provided by the embodiment of this application are similar to the autonomous driving test method of the previous embodiment, and will not be described again here.

Embodiments of the present application also provide a computer-readable storage medium. The computer-readable storage medium stores a computer program. When the computer program is executed by a processor, the processor causes the processor to implement the automatic driving test method provided in the above embodiments. A step of.

It should be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting.

It will also be understood that the term "and/or" as used in this application and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

The above are only specific embodiments of the present application, but the protection scope of the present application is not limited thereto. Any person familiar with the technical field can easily think of various equivalent methods within the technical scope disclosed in the present application. Modification or replacement, these modifications or replacements shall be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims (76)

1. An automated driving test method, the test method comprising:

   transmitting a preset test instruction to an unmanned aerial vehicle, wherein the test instruction is used for indicating the unmanned aerial vehicle to move in a traffic scene, and the unmanned aerial vehicle carries a model of a target object which moves along with the unmanned aerial vehicle in the traffic scene;

   Acquiring a running state of a vehicle to be tested in the traffic scene, wherein the vehicle to be tested can autonomously move in the traffic scene based on observation data of a model of the target object and a preset automatic driving algorithm;

   and generating a test result of the tested vehicle according to the running state of the tested vehicle.
2. The automated driving test method of claim 1, wherein the model of the target comprises one or more of: flexible board, flexible membrane, inflatable gasbag.
3. The automated driving test method of claim 2, wherein the airbag is capable of inflating or deflating when mounted on the drone.
4. The autopilot test method of claim 3 wherein the airbag is inflated by one or more of: the unmanned aerial vehicle is characterized in that the unmanned aerial vehicle is inflated by air flow generated by the blades during flight, by an inflation device on the air bag and by an external inflation device connected with the air bag.
5. The autopilot test method of claim 1 wherein the maximum weight of the model of the target is positively correlated with the area covered by the unmanned aerial vehicle's propeller disk.
6. The automated driving test method of claim 1, wherein the model of the target has an exterior shape similar to a traffic participant, the traffic participant comprising one or more of: motor vehicles, non-motor vehicles, pedestrians, animals.
7. The automated driving test method of claim 1, wherein the model of the object and/or the material of all or part of the outer surface of the drone is set according to the radar type of the vehicle under test.
8. The autopilot testing method of claim 7 wherein the outer surface is configured to detect signals for the radar that one or more of: diffuse reflection characteristics, refractive characteristics, absorption characteristics.
9. The method for testing the automatic driving according to claim 8, wherein the detection signal of the radar is any one of laser, millimeter wave and ultrasonic wave.
10. The automated driving test method of claim 1, wherein the model of the drone and the target are detachably connected.
11. The automated driving test method of claim 1, wherein the model of the drone and the target are rotatably connected.
12. The automated driving test method of claim 1, wherein all or part of the model of the object is automatically disconnectable from the unmanned aerial vehicle in a preset state.
13. The autopilot test method of claim 12 wherein the preset state includes: the tensile force between the model of the target object and the unmanned aerial vehicle is larger than a preset threshold value; and/or the tensile force direction between the model of the target object and the unmanned aerial vehicle is in a preset direction.
14. The automated driving test method of claim 12, wherein the automatically disconnecting the model of the object from the drone in a preset state is based on disconnecting at least two components of a connection between the model of the object and the drone.
15. The automated driving test method of claim 12, wherein the model of the target is connected to the unmanned aerial vehicle based on a connection that disconnects the model of the target from the unmanned aerial vehicle in response to a trigger instruction to enter the preset state.
16. The automated driving test method of claim 15, wherein the trigger instruction is generated by a sensor sensing a pull force between the model of the target object and the drone, the sensor being disposed on the model of the target object and/or the drone.
17. The autopilot testing method of claim 1 wherein the model of the target includes a plurality of components that are detachably connected.
18. The autopilot testing method of claim 17 wherein at least two of the plurality of components connect different drones.
19. The autopilot testing method of any one of claims 1-18 wherein the method further comprises:

    and when the running state of the tested vehicle does not meet the preset running state condition, controlling the unmanned aerial vehicle to execute a preset avoidance task so as to enable the model of the target object to avoid the tested vehicle.
20. The method for automatic driving test according to claim 19, wherein the controlling the unmanned aerial vehicle to perform a preset avoidance task to cause the model of the target object to avoid the vehicle under test includes:

    and controlling the unmanned aerial vehicle to adjust the motion state in the horizontal direction and/or the vertical direction so as to enable the model of the target object to avoid the tested vehicle.
21. The automated driving test method of claim 20, wherein the acceleration of the unmanned aerial vehicle in adjusting the state of motion in the vertical direction is determined from the weight of the model of the target.
22. The automated driving test method of claim 19, wherein the controlling the unmanned aerial vehicle to perform a preset avoidance task to cause the model of the target to avoid the vehicle under test comprises:

    and controlling the unmanned aerial vehicle to adjust the pose of the model of the target object so as to enable the bottom end of the model of the target object to be far away from the ground.
23. The automated driving test method of claim 19, wherein the controlling the unmanned aerial vehicle to perform a preset avoidance task to cause the model of the target to avoid the vehicle under test comprises:

    and controlling the unmanned aerial vehicle to execute a preset avoidance task so as to separate a plurality of parts which are detachably connected with the model of the target object from each other.
24. The automated driving test method of claim 23, wherein the number of parts located on the path of travel of the vehicle under test decreases after the plurality of parts are separated from one another.
25. The autopilot testing method of claim 23 wherein the plurality of components are separated from one another under the influence of a depressed wind field of unmanned aerial vehicle blades when the unmanned aerial vehicle performs the avoidance task and/or under the driving action of mechanical structure when the unmanned aerial vehicle performs the avoidance task.
26. The automated driving test method of claim 23, wherein controlling the unmanned aerial vehicle to perform a preset avoidance task to separate the detachably connected components of the model of the target from each other comprises:

    and controlling the unmanned aerial vehicles to be far away from the tested vehicle in different directions, so that the unmanned aerial vehicles are mutually separated from each other and are far away from the tested vehicle in different directions with a plurality of parts of the model of the target object.
27. The autopilot testing method of any one of claims 23-26 wherein at least two adjacent components of the plurality of components are connected by one or more of: the clamping and buckling, interference fit, magnetic attraction and adhesive connection.
28. The autopilot testing method of any one of claims 1-18 wherein the method further comprises:

    and when the running state of the tested vehicle does not meet the preset running state condition, disconnecting all or part of the model of the target object from the unmanned aerial vehicle.
29. The autopilot testing method of any one of claims 1-28 wherein the method further comprises:

    Controlling the operation of the environment working condition simulation device carried by the unmanned aerial vehicle, so that the environment working condition simulation device simulates one or more of the following environments: rainfall, dense fog, dust, illumination.
30. The automated driving test method of any one of claims 1-28, wherein the test instructions are for instructing the drone to move in a traffic scenario with a preset test trajectory.
31. The automated driving test method of any one of claims 1-28, wherein the generating a test result of the vehicle under test based on the driving state of the vehicle under test comprises:

    and generating a test result of the tested vehicle according to the movement state of the unmanned aerial vehicle and the running state of the tested vehicle.
32. The automated driving test method of any one of claims 1-28, wherein the test results of the vehicle under test comprise a data set that is a pre-set of the driving state of the vehicle under test.
33. The automated driving test method of any one of claims 19 to 28, wherein whether the running state of the vehicle under test satisfies the running state condition is determined from the running state of the vehicle under test obtained by one sampling period or from a trend of change in the running state of the vehicle under test obtained by a plurality of sampling periods.
34. The automated driving test method of claim 33, wherein the driving condition of the vehicle under test comprises at least one of:

    the method comprises the steps of measuring the motion parameters of a measured vehicle, observing information of the measured vehicle on a traffic scene, control information of the measured vehicle and the relative motion relation between the measured vehicle and other objects in the traffic scene.
35. The automated driving test method of claim 33, wherein whether the driving state of the vehicle under test satisfies the driving state condition is determined based on a relative positional relationship of the vehicle under test and a model of the target object.
36. The automated driving test method of claim 33, wherein whether the driving state of the vehicle under test satisfies the driving state condition is determined based on a preset type of evaluation index, the evaluation index being determined based on the driving state of the vehicle under test.
37. The automated driving test method of claim 33, wherein whether the driving state of the vehicle under test satisfies the driving state condition is determined from the obtained driving state of the vehicle under test and an expected driving state of the vehicle under test, the expected driving state being determined from the test instruction.
38. The automated driving test method of claim 33, wherein whether the driving state of the vehicle under test satisfies the driving state condition is determined based on whether the observed information of the vehicle under test for the traffic scene is consistent with the observed information of the model of the object for the traffic scene.
39. An automated driving test method for an unmanned aerial vehicle, the unmanned aerial vehicle being capable of carrying a model of a target object, the test method comprising:

    receiving a test instruction;

    and moving in a traffic scene according to the test instruction so that the model of the target object moves in the traffic scene along with the unmanned aerial vehicle.
40. The automated driving test method of claim 39, wherein the model of the drone and the target are detachably connected.
41. The automated driving test method of claim 39, wherein the model of the drone and the target are rotatably connected.
42. The automated driving test method of claim 39, wherein all or part of the model of the target object is automatically disconnected from the drone in a preset state.
43. The automated driving test method of claim 39, wherein the model of the target comprises a plurality of components, the plurality of components being detachably connected.
44. The automated driving test method of claim 43, wherein at least two of the plurality of components are connected to different drones.
45. The automated driving test method of any one of claims 39-44, wherein the method further comprises:

    and when the running state of the tested vehicle in the traffic scene does not meet the preset running state condition, executing a preset avoidance task so as to enable the model of the target object to avoid the tested vehicle.
46. The automated driving test method of claim 45, wherein performing a preset avoidance task to cause the model of the target to avoid the vehicle under test comprises:

    and adjusting the motion state in the horizontal direction and/or the vertical direction so as to enable the model of the target object to avoid the tested vehicle.
47. The automated driving test method of claim 45, wherein performing a preset avoidance task to cause the model of the target to avoid the vehicle under test comprises:

    And adjusting the pose of the model of the target object so as to enable the bottom end of the model of the target object to be far away from the ground.
48. The automated driving test method of claim 45, wherein performing a preset avoidance task to cause the model of the target to avoid the vehicle under test comprises:

    and executing a preset avoidance task so as to separate the detachably connected parts of the model of the target object from each other.
49. The automated driving test method of any one of claims 39-44, wherein the method further comprises:

    and when the running state of the tested vehicle in the traffic scene does not meet the preset running state condition, disconnecting all or part of the model of the target object.
50. An automated driving test method for a vehicle under test, the test method comprising:

    autonomous movement is performed in a traffic scene based on observation data of a model of a target object in the traffic scene and a preset automatic driving algorithm, wherein the traffic scene comprises an unmanned aerial vehicle and the model of the target object carried by the unmanned aerial vehicle, and the model of the target object moves in the traffic scene along with the unmanned aerial vehicle;

    Acquiring the running state of the tested vehicle;

    and sending the running state of the tested vehicle to an automatic driving testing device so that the automatic driving testing device generates a testing result of the tested vehicle according to the running state of the tested vehicle.
51. The automated driving test method of claim 50, wherein the driving status of the vehicle under test is further for: and when the automatic driving testing device determines that the running state of the tested vehicle does not meet the preset running state condition, controlling the unmanned aerial vehicle to execute a preset avoidance task so as to enable the model of the target object to avoid the tested vehicle.
52. The automated driving test method of claim 50, wherein the driving status of the vehicle under test is further for: and when the automatic driving testing device determines that the running state of the tested vehicle does not meet the preset running state condition, disconnecting all or part of the model of the target object from the unmanned aerial vehicle.
53. An autopilot testing apparatus comprising one or more processors operable individually or collectively to perform the steps of the autopilot testing method of any one of claims 1-38.
54. An unmanned aerial vehicle, comprising:

    the flight platform is used for flying;

    one or more processors, working individually or collectively, to perform the steps of the autopilot testing method of any one of claims 39-49.
55. A movable target, the movable target comprising:

    the drone of claim 54;

    the model of the target object can be connected to the unmanned aerial vehicle and moves in a traffic scene along with the unmanned aerial vehicle.
56. The movable object of claim 55, wherein the model of the object comprises one or more of: flexible board, flexible membrane, inflatable gasbag.
57. The movable target according to claim 56, wherein the airbag is capable of being inflated or deflated when mounted on the unmanned aerial vehicle.
58. The movable target according to claim 57, wherein the balloon is inflated by one or more of: the unmanned aerial vehicle is characterized in that the unmanned aerial vehicle is inflated by air flow generated by the blades during flight, by an inflation device on the air bag and by an external inflation device connected with the air bag.
59. The movable object of claim 55 wherein the maximum weight of the model of the object is positively correlated with the area covered by the unmanned aerial vehicle's paddles.
60. The movable object of claim 55 wherein the model of the object has an exterior shape similar to a traffic participant, the traffic participant comprising one or more of: motor vehicles, non-motor vehicles, pedestrians, animals.
61. The movable object of claim 55, wherein the model of the object and/or the material of all or part of the outer surface of the drone is set according to the radar type of the vehicle under test in the traffic scene.
62. The movable target of claim 61, wherein the outer surface is configured to detect signals for the radar in one or more of the following: diffuse reflection characteristics, refractive characteristics, absorption characteristics.
63. The movable target of claim 62, wherein the radar detection signal is any one of laser, millimeter wave, and ultrasonic.
64. The movable object of claim 55, wherein the model of the object is detachably connectable to the drone.
65. The movable object of claim 55 wherein the model of the object is capable of being rotatably coupled to the unmanned aerial vehicle.
66. The movable object of claim 55, wherein all or part of the model of the object is automatically disconnected from the drone in a preset state.
67. The movable target of claim 66, wherein the predetermined state comprises: the tensile force between the model of the target object and the unmanned aerial vehicle is larger than a preset threshold value; and/or the tensile force direction between the model of the target object and the unmanned aerial vehicle is in a preset direction.
68. The movable object according to claim 66, wherein the model of the object is automatically disconnected from the unmanned aerial vehicle in a predetermined state based on the disconnection of at least two parts of a connection between the model of the object and the unmanned aerial vehicle.
69. The movable object according to claim 66, wherein the model of the object is connected to the unmanned aerial vehicle based on a connection that disconnects the model of the object from the unmanned aerial vehicle in response to a trigger instruction to enter the preset state.
70. The movable target of claim 69, wherein the trigger instruction is generated by a sensor sensing a pulling force between a model of the target and the drone, the sensor being disposed on the model of the target and/or the drone.
71. The movable object of claim 55 wherein the model of the object comprises a plurality of components, the plurality of components being detachably connected.
72. The movable target of claim 71, wherein at least two of the plurality of components are connected to different drones.
73. A vehicle, characterized by comprising:

    a vehicle platform;

    one or more processors, working individually or collectively, to perform the steps of the autopilot testing method of any one of claims 50-52.
74. An autopilot test system comprising:

    unmanned plane;

    the model of the target object can be carried on the unmanned aerial vehicle and moves in a traffic scene along with the unmanned aerial vehicle;

    the detected vehicle can autonomously move in the traffic scene based on the observation data of the model of the target object and a preset automatic driving algorithm;

    the autopilot test unit of claim 53.
75. The autopilot testing system of claim 74 wherein the autopilot testing arrangement is disposed on the drone.
76. A computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to implement:

    An autopilot testing method as claimed in any one of claims 1 to 38; and/or

    An autopilot testing method as claimed in any one of claims 39 to 49; and/or

    An autopilot testing method as claimed in any one of claims 50 to 52.