Key scene recognition method, device, equipment and storage medium

Abstract

The present invention belongs to the field of intelligent driving technology, and discloses a key scene recognition method, device, equipment and storage medium. The method comprises: obtaining road test data and user driving data; analyzing the road test data and user driving data according to a preset environment model to obtain scene evaluation information; evaluating the scene evaluation information to identify key scenes. Through the above method, it is possible to efficiently screen out key scenes based on road test data and user driving data to guide the functional design of the vehicle, improve the functional scene coverage and user experience.

Description

Technical Field

[0001] The present invention relates to the field of intelligent driving technology, and in particular to a key scene recognition method, device, equipment and storage medium. Background Art

[0002] In recent years, intelligent network technology has developed rapidly, and more complex autonomous driving functions are getting closer to consumers. However, the design domain of the function is difficult to cover the consumer's usage scenarios, which limits the use of the function and brings potential safety risks. Efficiently identifying key scenarios in huge vehicle data has become a technical problem that needs to be solved urgently.

[0003] The above contents are only used to assist in understanding the technical solution of the present invention and do not constitute an admission that the above contents are prior art. Summary of the invention

[0004] The main purpose of the present invention is to provide a key scene recognition method, device, equipment and storage medium, aiming to solve the technical problem of how to recognize key scenes in the prior art.

[0005] To achieve the above object, the present invention provides a key scene recognition method, the method comprising the following steps:

[0006] Obtain road test data and user driving data;

[0007] Analyze the road test data and user driving data according to the preset environment model to obtain scenario evaluation information;

[0008] The scenario assessment information is evaluated to identify key scenarios.

[0009] Optionally, analyzing the road test data and the user driving data according to the preset environment model to obtain the scenario evaluation information includes:

[0010] Determine driver model parameters according to the road test data and the user driving data, and construct a driver behavior model according to the driver model parameters;

[0011] Determine environmental parameters according to the road test data and the user driving data, and construct an environmental factor model according to the environmental parameters;

[0012] A preset scenario assessment model is input according to the driver behavior model and the environmental factor model to obtain scenario assessment information.

[0013] Optionally, constructing a driver behavior model according to the driver model parameters includes:

[0014] Determining vehicle motion state information and vehicle driving information according to the driver model parameters;

[0015] A driver behavior model is determined according to the vehicle motion state information and the vehicle driving information.

[0016] Optionally, determining the environmental factor model according to the environmental parameters includes:

[0017] Determine surrounding vehicle motion parameters, pedestrian parameters, and traffic instruction information parameters according to the environmental parameters;

[0018] An environmental element model is generated according to the vehicle motion parameters, pedestrian parameters and traffic instruction information parameters.

[0019] Optionally, before obtaining scene assessment information by inputting a preset scene assessment model according to the driver behavior model and the environmental factor model, the method includes:

[0020] Determining driving related parameters, wherein the driving related parameters include driving safety parameters, driving function parameters, and comfort evaluation parameters;

[0021] Determining a reward and punishment function according to the driving-related parameters;

[0022] A scenario evaluation model is determined according to the reward and punishment function.

[0023] Optionally, determining a reward and punishment function according to the driving-related parameters includes:

[0024] Determining a driving habit weight value and a driving environment weight value according to the driving related parameters;

[0025] The preset initial reward and punishment function is adjusted according to the driving habit weight value and the driving environment weight value to obtain a reward and punishment function.

[0026] Optionally, analyzing the road test data and the user driving data according to the preset environment model to obtain the scenario evaluation information includes:

[0027] Performing data slicing and clustering analysis on the road test data and user driving data to determine static parameters;

[0028] The static parameters are input into a preset environment model to determine scene evaluation information.

[0029] In addition, to achieve the above-mentioned purpose, the present invention further proposes a key scene recognition device, the key scene recognition device comprising:

[0030] An acquisition module is used to acquire road test data and user driving data;

[0031] A processing module is used to analyze the road test data and the user driving data according to a preset environment model to obtain scenario evaluation information;

[0032] The processing module is further used to evaluate the scene evaluation information and identify key scenes.

[0033] In addition, to achieve the above-mentioned purpose, the present invention also proposes a key scene recognition device, which includes: a memory, a processor, and a key scene recognition program stored in the memory and executable on the processor, and the key scene recognition program is configured to implement the steps of the key scene recognition method described above.

[0034] In addition, to achieve the above-mentioned purpose, the present invention further proposes a storage medium, on which a key scene recognition program is stored, and when the key scene recognition program is executed by a processor, the steps of the key scene recognition method described above are implemented.

[0035] The present invention obtains road test data and user driving data; analyzes the road test data and user driving data according to a preset environment model to obtain scene evaluation information; and evaluates the scene evaluation information to identify key scenes. Through the above method, it is achieved to evaluate driving scenes and select more critical scenes based on test data such as road test data and user driving data to guide the functional design of the vehicle, thereby improving the comprehensiveness of vehicle design. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a structural diagram of a key scene recognition device in a hardware operating environment involved in an embodiment of the present invention;

[0037] Figure 2 It is a flowchart of a first embodiment of a key scene identification method of the present invention;

[0038] Figure 3 It is a flowchart of a second embodiment of a key scene identification method of the present invention;

[0039] Figure 4 It is a structural block diagram of the first embodiment of the key scene recognition device of the present invention.

[0040] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0041] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0042] Reference Figure 1 , Figure 1 It is a schematic diagram of the structure of a key scene recognition device in the hardware operating environment involved in the embodiment of the present invention.

[0043] like Figure 1As shown, the key scene recognition device may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Among them, the communication bus 1002 is used to realize the connection and communication between these components. The user interface 1003 may include a display screen (Display), an input unit such as a keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (such as a wireless fidelity (Wireless-Fidelity, Wi-Fi) interface). The memory 1005 may be a high-speed random access memory (Random Access Memory, RAM) memory, or a stable non-volatile memory (Non-Volatile Memory, NVM), such as a disk memory. The memory 1005 may also be a storage device independent of the aforementioned processor 1001.

[0044] Those skilled in the art will understand that Figure 1 The structure shown in the figure does not constitute a limitation on the key scene recognition device, and may include more or less components than those shown in the figure, or combine certain components, or arrange the components differently.

[0045] like Figure 1 As shown, the memory 1005 as a storage medium may include an operating system, a network communication module, a user interface module, and a key scene recognition program.

[0046] exist Figure 1 In the key scene recognition device shown, the network interface 1004 is mainly used for data communication with the network server; the user interface 1003 is mainly used for data interaction with the user; the processor 1001 and the memory 1005 in the key scene recognition device of the present invention can be set in the key scene recognition device, and the key scene recognition device calls the key scene recognition program stored in the memory 1005 through the processor 1001, and executes the key scene recognition method provided by the embodiment of the present invention.

[0047] The embodiment of the present invention provides a key scene recognition method, referring to Figure 2 , Figure 2 The figure is a flow chart of a first embodiment of a key scene recognition method of the present invention.

[0048] In this embodiment, the key scene recognition method includes the following steps:

[0049] Step S10: Acquire road test data and user driving data.

[0050] It should be noted that the executor of this embodiment is a smart terminal, which may be a vehicle controller, an on-board computer, or other devices with the same or similar functions as the vehicle controller. This embodiment does not limit this and only takes the vehicle controller as an example for explanation.

[0051] It should be noted that this embodiment is applied to the process of identifying key scenes through test data, identifying key scenes to ensure that the safety and performance of the autonomous driving system in various situations can be fully covered. Key scenes can be, for example: complex intersections (such as multi-lane, multi-lane turning, non-standard shaped intersections, etc.), highway lane changes, emergency response (such as obstacles suddenly appearing in front, pedestrians suddenly crossing the road, etc.), severe weather conditions (such as driving scenes in severe weather conditions such as rain, snow, fog, etc., which put forward higher requirements on the perception and control capabilities of the autonomous driving system, and require the system to be able to effectively deal with problems such as reduced visibility and slippery roads), complex road structures (including mountain bends, narrow roads, complex terrain, etc.) and crowded areas (such as city centers, commercial areas, etc.) How to efficiently, comprehensively and accurately find key scenes is the problem of the expected outcome of this embodiment.

[0052] It is understood that road test data generally refers to data collected when testing the performance of a vehicle under actual road conditions during the design, development and testing of the vehicle. These data may include the output of vehicle sensors (such as acceleration, steering angle, braking force, etc.), records of the vehicle electronic control unit (ECU), images or scanning data of vehicle cameras or radars, etc. Road test data is very important for evaluating the vehicle's handling, stability, acceleration, fuel efficiency, emissions, etc., and can be used to optimize vehicle design, adjust vehicle parameters, and verify the performance of the vehicle under actual road conditions. On the other hand, user driving data refers to data collected from the driver's perspective, recording the driver's behavior and the status of the vehicle during driving. These data are usually collected through on-board sensors, smartphone applications or specific data recording devices, and may include vehicle speed, acceleration, braking conditions, turning angles, driving routes, driving time, in-vehicle environment (such as temperature, humidity), and the driver's physiological indicators (such as heart rate, fatigue level). User driving data is very important for understanding the driver's driving habits, behavior patterns, route preferences, etc., and can be used to provide personalized driving suggestions, driving behavior analysis, insurance pricing, vehicle maintenance reminders and other services. In general, road test data is mainly used in the vehicle development and testing stages to evaluate the performance and safety of the vehicle, while user driving data is mainly used to understand driver behavior and provide personalized services.

[0053] Step S20: Analyze the road test data and user driving data according to the preset environment model to obtain scenario evaluation information.

[0054] It should be noted that the road test data and user driving data are analyzed according to the preset environment model. Since the road test data can determine the vehicle status and various information of the vehicle's current environment, it can be combined with the user operation and user reaction in the corresponding user driving data to determine whether the current moment has a significant impact on the user's operating experience and vehicle safety, so that evaluations in different scenarios can be obtained, thereby determining the scenario evaluation information. Specifically, for example, the collected data can be cleaned and preprocessed, including removing noise, filling missing values, calibrating the data, etc., to ensure the accuracy and availability of the data, and further using the preprocessed data, through algorithms and models to identify key scenarios that have a greater impact on safety and comfort, such as emergency braking, sudden acceleration, highway cruising, urban congestion, curve driving, etc., and machine learning, deep learning and other technologies can be used to train the model for scenario classification. Feature extraction is performed on the identified key scenarios, and key indicators related to the scenarios are extracted, such as braking distance, acceleration, vehicle speed, braking frequency, steering angle, etc. Analyze and evaluate the extracted scene features, compare the data performance in different scenes, find out the performance differences and potential problems, and propose improvement suggestions. Statistical analysis, data visualization and other methods can be used to visualize and analyze the scene data to intuitively understand the performance and driving behavior of the vehicle. This step can break the original scenes determined by expert experience and identify some key scenes that were not originally noticed.

[0055] In some embodiments, data slicing and clustering analysis are performed on the road test data and user driving data to determine static parameters; and the static parameters are input into a preset environmental model to determine scenario evaluation information.

[0056] It is understandable that the data slicing and clustering analysis of the road test data and user driving data is mainly to further determine the characteristic parameters of the vehicle. By performing data slicing and clustering analysis on the road test data, the static parameter differences between different vehicle models or vehicle configurations, the static parameters under different driving behavior modes, the static parameters under different road conditions, etc. can be determined, such as driver behavior model parameters, such as maximum angular velocity, maximum acceleration, maximum jerk, etc., vehicle performance parameters, relatively static road environment, meteorological conditions, etc. These parameters will not change significantly over time and space.

[0057] Step S30: Evaluate the scene evaluation information and identify key scenes.

[0058] It should be noted that after extracting the scene evaluation information, the importance of the scene can be determined as the basis for extracting key scenes. For example, the scene evaluation information can be directly the scene importance value, or it can be scores in multiple dimensions, such as safety impact score, comfort impact score, etc.

[0059] This embodiment obtains road test data and user driving data; analyzes the road test data and user driving data according to a preset environment model to obtain scene evaluation information; and evaluates the scene evaluation information to identify key scenes. Through the above method, it is achieved to evaluate driving scenes and select more critical scenes based on test data such as road test data and user driving data to guide the functional design of the vehicle, thereby improving the comprehensiveness of vehicle design.

[0060] refer to Figure 3 , Figure 3 The figure is a flow chart of a second embodiment of a key scene recognition method according to the present invention.

[0061] Based on the first embodiment, the key scene recognition method of this embodiment further includes, in step S20:

[0062] Step S21: determining driver model parameters according to the road test data and the user driving data, and constructing a driver behavior model according to the driver model parameters.

[0063] It should be noted that the road test data and the user driving data are subjected to data slicing and cluster analysis to determine static parameters; the static parameters are input into the preset environment model to determine the scene evaluation information. Specifically, the static sampling parameters can be obtained in the above manner, for example: according to the road test data and the user driving data, the static parameters are extracted through data slicing and cluster analysis, including the driver behavior model parameters, such as the maximum angular velocity, the maximum acceleration, the maximum jerk, etc., the vehicle performance parameters, the relatively static road environment, the meteorological conditions and other parameters, And the corresponding conditional probability distribution; static parameter sampling. According to the obtained conditional probability distribution, random sampling is performed. Due to the possible correlation of parameters and the difficulty of sampling complex probability distributions, the Markov-Monte Carlo method is used here for parameter sampling. The specific process is as follows: 1. Any Markov transition matrix Q and parameter probability distribution π(x), state transition threshold n1, number of samples n2; sample from any simple probability distribution to obtain the initial parameter sample: The cycle is repeated from t = 0 to n1 + n2 - 1. The process is as follows: From the conditional probability distribution Get sample x x ; Sampling from uniform distribution uniform[0,1] if Then accept xx , that is, to transfer the state, Otherwise, the transfer is not accepted, i.e.: x t+1 =x t . j represents the state x x , that is, the new state, i is the old state x t .

[0064] It should be noted that a driver behavior model can be constructed through road test data and user driving data, and the driver's behavior and the driver's control over the vehicle can be simulated through the model.

[0065] In some embodiments, vehicle motion state information and vehicle driving information are determined based on the driver model parameters; and a driver behavior model is determined based on the vehicle motion state information and vehicle driving information.

[0066] It should be noted that setting the driver behavior model includes setting the original vehicle motion state and the main vehicle driving model. The relevant parameters of the driver model include the acceleration coefficient, the main vehicle speed, the expected driving speed of the main vehicle, the minimum safety distance, etc. The aforementioned sampling parameters can be used as the input of the driver behavior model.

[0067] Step S22: determining environmental parameters according to the road test data and the user driving data, and constructing an environmental factor model according to the environmental parameters.

[0068] It should be noted that the environmental factor model is a model used to simulate environmental conditions. Environmental parameters are determined based on road test data and user driving data, and an environmental factor model is constructed. First, a large amount of road test data and user driving data needs to be collected, and the collected data needs to be processed and cleaned to ensure data quality and accuracy. Data interpolation, denoising and other processing may be required. Features are extracted from the processed data, which may include vehicle motion status, environmental conditions, driving behavior, etc. Environmental parameters such as road friction coefficient, traffic density, road slope, etc. are determined using the extracted features and other information. Based on the determined environmental parameters, an environmental factor model is constructed. This can be a physical model, a statistical model or a machine learning model that describes the impact of the environment on vehicle travel and driving behavior. Through the above steps, environmental parameters can be determined based on road test data and user driving data, and an environmental factor model can be constructed.

[0069] In some embodiments, surrounding vehicle motion parameters, pedestrian parameters and traffic indication information parameters are determined based on the environmental parameters; and an environmental element model is generated based on the vehicle motion parameters, pedestrian parameters and traffic indication information parameters.

[0070] Specifically, the dynamic model of environmental elements includes pedestrians, other driving vehicles, traffic signs, etc. The generation of dynamic scenes is expressed as a Markov decision process and solved using a deep learning algorithm. Given an initial state (s0), the observation value of the state (p) is obtained, and an action (a) is taken. In addition, sensor errors are considered and an observation history set is constructed. When sensor data is wrong, it is randomly extracted from the history set.

[0071] Step S23: inputting a preset scenario assessment model according to the driver behavior model and the environmental factor model to obtain scenario assessment information.

[0072] For the preset scenario evaluation model, we can combine the driver behavior model and the environmental element model, input specific scenario information, and then evaluate the safety, adaptability or other indicators of the scenario. First, the preset scenario needs to be clearly defined, including the vehicle's location, road type, traffic conditions, weather conditions, etc. Input the scenario information into the driver behavior model to simulate the driver's behavior in the scenario. This may involve the driver's decision-making process, operating behavior, etc. Input the scenario information into the environmental element model to simulate the impact of the environment on vehicle driving, including road conditions, road speed limits, traffic flow, etc. Based on the output of the driver behavior model and the environmental element model, evaluate the safety, comfort or other indicators of the scenario. This may involve analysis of driver behavior, simulation of vehicle dynamics, etc. Output the evaluation results in a visual or other form to obtain scenario evaluation information.

[0073] In some embodiments, driving related parameters are determined, wherein the driving related parameters include driving safety parameters, driving function parameters, and comfort evaluation parameters; a reward and punishment function is determined based on the driving related parameters; and a scenario evaluation model is determined based on the reward and punishment function.

[0074] Specifically, setting the reward and punishment function according to the driving safety parameters, driving function parameters and comfort evaluation parameters can more easily guide the improvement of the autonomous driving function from the perspective of safety and comfort. According to these parameters, the reward and punishment function can be defined to evaluate the pros and cons of driving in different scenarios. The design of the reward and punishment function needs to take into account the trade-offs between safety, functionality and comfort, as well as the weight relationship between different parameters. For example, for driving safety parameters, a reward function can be defined so that safer driving behaviors are rewarded, such as avoiding collisions or reducing braking distances. For comfort parameters, a penalty function can be defined to penalize large acceleration changes or vehicle vibration levels. For driving function parameters, rewards and penalties can be given based on factors such as system response speed and path planning accuracy. Then, based on the defined reward and punishment function, a scenario evaluation model can be established to evaluate driving performance in different driving scenarios. This model can be implemented using machine learning methods, optimization algorithms or simulation techniques. The scenario evaluation model can help optimize driving strategies and system design, and improve driving safety, functionality and comfort.

[0075] In some embodiments, a driving habit weight value and a driving environment weight value are determined according to the driving-related parameters; a preset initial reward and punishment function is adjusted according to the driving habit weight value and the driving environment weight value to obtain a reward and punishment function.

[0076] In the specific implementation, the scenario criticality evaluation model is set up, and the reward and punishment function is set based on relevant physical parameters such as driving safety, function, and comfort. It is necessary to consider adjusting its weight according to actual driving habits and driving environment in order to identify the corresponding critical scenarios. Furthermore, in the simulation or model analysis process, it is necessary to solve the scenario to determine the corresponding critical scenario, for example: select the Monte Carlo tree search method to solve the MDP problem. Select, starting from the initial state, according to UCB (UpperConfidenceBound), select the action with the largest UCB. V i is the average value of the node, N is the total number of explorations, n i is the number of explorations of the current node. For the above dynamic environment factors, explore the action with the largest UCB, where That is, critical evaluation, setting reward values ​​based on driving safety, function, comfort, etc. It can be defined as V z (s)=R(s,π(s))+γΣ s ·T(s′|s,π(s))V z (s′) where V z(s) is the expected reward value of taking action π(s) from state s, R(s,π(s)) is the reward function, γ is the future reward weight, and T(s′|s,π(s)) is the state transition matrix from state s to state s′. Expansion: expand the leaf nodes to generate new child nodes; simulation: simulate the new nodes, start from the expanded nodes, run a complete decision-making process, set the reward threshold, and calculate the success rate of the simulation accordingly; back propagation: trace back the results of the simulation to find the corresponding actions and key scene sets.

[0077] This embodiment determines the driver model parameters according to the road test data and the user driving data, and constructs the driver behavior model according to the driver model parameters; determines the environmental parameters according to the road test data and the user driving data, and constructs the environmental factor model according to the environmental parameters; and inputs the preset scene evaluation model according to the driver behavior model and the environmental factor model to obtain the scene evaluation information. By setting the environmental factor model and the driver behavior model, the importance of the scene is evaluated, and the accuracy and comprehensiveness of the key scene extraction are improved.

[0078] In addition, an embodiment of the present invention further provides a storage medium, on which a key scene recognition program is stored. When the key scene recognition program is executed by a processor, the steps of the key scene recognition method described above are implemented.

[0079] Reference Figure 4 , Figure 4 It is a structural block diagram of the first embodiment of the key scene recognition device of the present invention.

[0080] like Figure 4 As shown, the key scene recognition device proposed in the embodiment of the present invention includes:

[0081] An acquisition module 10 is used to acquire road test data and user driving data;

[0082] The processing module 20 is used to analyze the road test data and the user driving data according to the preset environment model to obtain the scene evaluation information;

[0083] The processing module 20 is further configured to evaluate the scene evaluation information and identify key scenes.

[0084] It should be understood that the above is only an example and does not constitute any limitation on the technical solution of the present invention. In specific applications, technicians in this field can make settings as needed, and the present invention does not limit this.

[0085] In this embodiment, the acquisition module 10 acquires the road test data and the user driving data; the processing module 20 analyzes the road test data and the user driving data according to the preset environment model to obtain the scene evaluation information; the processing module 20 evaluates the scene evaluation information to identify the key scenes. In the above manner, it is achieved to evaluate the driving scenes and select the more critical scenes according to the test data such as the road test data and the user driving data, so as to guide the functional design of the vehicle and improve the comprehensiveness of the vehicle design.

[0086] In some embodiments, the processing module 20 is further used to determine the driver model parameters according to the road test data and the user driving data, and to construct the driver behavior model according to the driver model parameters;

[0087] Determine environmental parameters according to the road test data and the user driving data, and construct an environmental factor model according to the environmental parameters;

[0088] A preset scenario assessment model is input according to the driver behavior model and the environmental factor model to obtain scenario assessment information.

[0089] In some embodiments, the processing module 20 is further used to determine vehicle motion state information and vehicle driving information according to the driver model parameters;

[0090] A driver behavior model is determined according to the vehicle motion state information and the vehicle driving information.

[0091] In some embodiments, the processing module 20 is also used to

[0092] Determine surrounding vehicle motion parameters, pedestrian parameters, and traffic instruction information parameters according to the environmental parameters;

[0093] An environmental element model is generated according to the vehicle motion parameters, pedestrian parameters and traffic instruction information parameters.

[0094] In some embodiments, the processing module 20 is further used to determine driving related parameters, wherein the driving related parameters include driving safety parameters, driving function parameters, and comfort evaluation parameters;

[0095] Determining a reward and punishment function according to the driving-related parameters;

[0096] A scenario evaluation model is determined according to the reward and punishment function.

[0097] In some embodiments, the processing module 20 is further used to determine a driving habit weight value and a driving environment weight value according to the driving-related parameters;

[0098] The preset initial reward and punishment function is adjusted according to the driving habit weight value and the driving environment weight value to obtain a reward and punishment function.

[0099] In some embodiments, the processing module 20 is further used to perform data slicing and cluster analysis on the road test data and the user driving data to determine static parameters;

[0100] The static parameters are input into a preset environment model to determine scene evaluation information.

[0101] It should be noted that the workflow described above is merely illustrative and does not limit the scope of protection of the present invention. In practical applications, technicians in this field can select part or all of them according to actual needs to achieve the purpose of the present embodiment, and no limitation is made here.

[0102] In addition, for technical details not fully described in this embodiment, reference can be made to the key scene recognition method provided in any embodiment of the present invention, and will not be repeated here.

[0103] In addition, it should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or system including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or system including the element.

[0104] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0105] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as a read-only memory (ROM) / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0106] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A key scene recognition method, characterized in that: The key scene recognition method comprises: Obtain road test data and user driving data; Analyze the road test data and user driving data according to the preset environment model to obtain scenario evaluation information; Evaluating the scenario evaluation information to identify key scenarios; The analyzing the road test data and the user driving data according to the preset environment model to obtain the scene evaluation information includes: Determine driver model parameters according to the road test data and the user driving data, and construct a driver behavior model according to the driver model parameters; Determine environmental parameters according to the road test data and the user driving data, and construct an environmental factor model according to the environmental parameters; Inputting a preset scenario assessment model according to the driver behavior model and the environmental factor model to obtain scenario assessment information; Before obtaining scene evaluation information by inputting a preset scene evaluation model according to the driver behavior model and the environmental factor model, the method includes: Determining driving related parameters, wherein the driving related parameters include driving safety parameters, driving function parameters, and comfort evaluation parameters; Determining a reward and punishment function according to the driving-related parameters; A scenario evaluation model is determined according to the reward and punishment function.

2. The method according to claim 1, characterized in that The step of constructing a driver behavior model according to the driver model parameters comprises: Determining vehicle motion state information and vehicle driving information according to the driver model parameters; A driver behavior model is determined according to the vehicle motion state information and the vehicle driving information.

3. The method according to claim 1, characterized in that Determining the environmental factor model according to the environmental parameters includes: Determine surrounding vehicle motion parameters, pedestrian parameters, and traffic instruction information parameters according to the environmental parameters; An environmental element model is generated according to the vehicle motion parameters, pedestrian parameters and traffic instruction information parameters.

4. The method according to claim 1, characterized in that Determining the reward and punishment function according to the driving related parameters includes: Determining a driving habit weight value and a driving environment weight value according to the driving related parameters; The preset initial reward and punishment function is adjusted according to the driving habit weight value and the driving environment weight value to obtain a reward and punishment function.

5. The method according to claim 1, characterized in that The analyzing the road test data and the user driving data according to the preset environment model to obtain the scene evaluation information includes: Performing data slicing and clustering analysis on the road test data and user driving data to determine static parameters; The static parameters are input into a preset environment model to determine scene evaluation information.

6. A key scene recognition device, characterized in that: The key scene recognition device comprises: An acquisition module is used to acquire road test data and user driving data; A processing module is used to analyze the road test data and the user driving data according to a preset environment model to obtain scenario evaluation information; The processing module is further used to evaluate the scene evaluation information and identify key scenes; The processing module is further used to determine the driver model parameters according to the road test data and the user driving data, and to construct the driver behavior model according to the driver model parameters; Determine environmental parameters according to the road test data and the user driving data, and construct an environmental factor model according to the environmental parameters; Inputting a preset scenario assessment model according to the driver behavior model and the environmental factor model to obtain scenario assessment information; The processing module is further used to determine driving related parameters, wherein the driving related parameters include driving safety parameters, driving function parameters and comfort evaluation parameters; Determining a reward and punishment function according to the driving-related parameters; A scenario evaluation model is determined according to the reward and punishment function.

7. A key scene recognition device, characterized in that: The device comprises: a memory, a processor, and a key scene recognition program stored in the memory and executable on the processor, wherein the key scene recognition program is configured to implement the steps of the key scene recognition method according to any one of claims 1 to 5.

8. A storage medium, characterized in that: The storage medium stores a key scene recognition program, and when the key scene recognition program is executed by the processor, the steps of the key scene recognition method according to any one of claims 1 to 5 are implemented.

Images