A human-machine interaction method and system based on an autonomous vehicle and a medium

Abstract

The application relates to a human-computer interaction method and system based on an automatic driving vehicle and a medium, the method comprising the following steps: L1. The automatic driving vehicle travels on a road, real-time image data information of the road is acquired based on a vehicle-mounted camera, point cloud data information of the road is acquired based on a vehicle-mounted laser radar, and preprocessing is performed to obtain processed image and point cloud data information of the road; L2. Based on the processed image and point cloud data information of the road, a space-time graph of the road is constructed, feature extraction of pedestrians is performed to obtain feature space-time graph data information of the pedestrians, a pedestrian relationship kernel function trajectory algorithm is used to calculate the trajectory of the pedestrians, and trajectory data information of the pedestrians is output. The application can not only more accurately predict the trajectory of the pedestrians, but also has a more reasonable human-computer interaction mode and does not produce noise pollution, so that the operation efficiency of the unmanned driving vehicle is improved.

Description

Technical Field

[0001] This invention relates to the field of autonomous driving technology, and in particular to a human-computer interaction method, system and medium based on autonomous vehicles. Background Technology

[0002] In urban transportation systems, the interaction between pedestrians and vehicles is crucial. Current technologies that use vehicle horns to remind road users have drawbacks such as noise pollution, unfriendly interaction methods, limited alerting options, and a lack of hazard classification for road users. How to provide friendly alerts to road users without generating noise pollution has become an urgent problem to be solved.

[0003] In the prior art, a patent (application number: 202110882258.9) discloses a method, device, electronic device, and storage medium for predicting pedestrian trajectories, including: acquiring observation trajectory information of at least one pedestrian in a scene; converting the observation trajectory information of each pedestrian into self-perspective trajectory information from each pedestrian's own viewpoint; acquiring the movement trend characteristics of each pedestrian based on their self-perspective trajectory information, and acquiring the interaction characteristics between each pedestrian and other pedestrians; generating future position information from each pedestrian's own viewpoint based at least on the movement trend characteristics and the interaction characteristics between each pedestrian and other pedestrians; and generating at least one future trajectory from each pedestrian's own viewpoint based at least on their future position information from their own viewpoint, and converting the self-perspective future trajectory into a future trajectory in a world coordinate system. However, the prediction of pedestrian trajectories is not yet perfect and needs further development. Summary of the Invention

[0004] In view of the above problems, the present invention provides a human-computer interaction method, system and medium based on autonomous vehicles, which can not only predict pedestrian trajectories more accurately, but also make the human-computer interaction method more reasonable and will not generate noise pollution, thereby improving the operating efficiency of autonomous vehicles.

[0005] To achieve the above and other related objectives, the present invention provides the following technical solution:

[0006] A human-machine interaction method based on autonomous vehicles, the method comprising:

[0007] L1. Autonomous vehicles drive on the road, acquiring real-time image data of the road based on onboard cameras and real-time point cloud data of the road based on onboard LiDAR, and preprocessing the data to obtain processed image and point cloud data of the road.

[0008] L2. Based on the processed road image and point cloud data, construct a spatiotemporal map of the road, extract pedestrian features to obtain pedestrian feature spatiotemporal map data, use the pedestrian relationship kernel function trajectory algorithm to calculate the pedestrian trajectory, and output the pedestrian trajectory data.

[0009] L3. Input the pedestrian trajectory data information into the trained spatiotemporal graph convolutional neural network model for feature extraction to obtain the feature data information of the pedestrian trajectory. Then, input the feature data information of the pedestrian trajectory into the trained temporal extrapolator convolutional neural network model to predict the pedestrian trajectory and output the predicted pedestrian trajectory data information.

[0010] L4. Based on the predicted pedestrian trajectory data, a safe distance interaction algorithm is used to calculate the distance between pedestrians and vehicles, and obtain safe interaction distance data between pedestrians and vehicles;

[0011] L5. Based on the safe interaction distance data between pedestrians and vehicles, the interaction volume of the vehicle is adjusted using an interaction volume and distance control algorithm, and the interaction volume data of pedestrians and vehicles is output.

[0012] Furthermore, in step L2, the calculation of the pedestrian trajectory using the pedestrian relationship kernel function trajectory algorithm includes:

[0013] L21. Based on the characteristic spatiotemporal graph data information of the pedestrians, characterize the relationship between each pedestrian and output the relationship characterization data information of the pedestrians;

[0014] L22. Based on the pedestrian relationship representation data, establish a pedestrian relationship kernel function Q.

[0015] ,

[0016] i≠j,

[0017] Where α is a constant parameter matrix, β is an inner product scaling matrix, and x i Let x be the i-th pedestrian relationship matrix. j Let ω be the j-th pedestrian relationship matrix, and let ω be the pedestrian simulation control parameter, to obtain pedestrian relationship data information;

[0018] L23. Based on the pedestrian relationship data, a cubic spline function is used to fit the pedestrian trajectory to obtain the pedestrian trajectory data.

[0019] Furthermore, in step L23, fitting the pedestrian's trajectory using a cubic spline function includes:

[0020] L231. Based on the pedestrian relationship data, establish a cubic spline fitting function F for the pedestrian trajectory.

[0021] F=(Q×λ3)x 3 +(Q×λ2)x 2 +(Q×λ1)x+(Q×λ0),

[0022] Where Q represents pedestrian relationship data, λ0, λ1, λ2 and λ3 are the coefficient matrices of the cubic spline function, and x is the independent variable;

[0023] L232. Based on the cubic spline fitting function F of the pedestrian trajectory, the pedestrian trajectory is calculated to obtain the pedestrian trajectory data information.

[0024] Furthermore, in step L3, the trained spatiotemporal graph convolutional neural network model is obtained by dividing the pedestrian trajectory data information into a training dataset and a test training dataset, and inputting the training dataset into the spatiotemporal graph convolutional neural network model for training and learning, determining the network parameters, and obtaining the trained spatiotemporal graph convolutional neural network model.

[0025] Furthermore, the spatiotemporal graph convolutional neural network model includes a spatiotemporal graph convolutional input layer, a spatiotemporal graph convolutional kernel function, and a spatiotemporal graph convolutional output layer, wherein the spatiotemporal graph convolutional kernel function is G.

[0026] ,

[0027] Where A is the weight matrix, (x,y,z) are the pedestrian trajectory coordinates, and η0, η1 and η2 are the learning factor parameters of the corresponding pedestrian trajectory coordinates.

[0028] Furthermore, the trained temporal extrapolator convolutional neural network model is obtained by dividing the feature data information of the pedestrian's trajectory into a training dataset and a test training dataset, and inputting the training dataset into the temporal extrapolator convolutional neural network model for training and learning, determining the network parameters, and obtaining the trained temporal extrapolator convolutional neural network model.

[0029] Furthermore, the temporal extrapolator convolutional neural network model includes a temporal extrapolator input layer, a temporal extrapolator kernel function, and a temporal extrapolator kernel function H.

[0030] ,

[0031] ,

[0032] Where f is the observation function, X i Let X be the feature data information of the i-th pedestrian's trajectory, and O be the feature data information of the pedestrian's trajectory.i Let X be the Lagrange basis function for the i-th pedestrian. j Let n be the feature data information of the trajectory of the j-th pedestrian, and n be the total number of samples of the feature data of the pedestrian's trajectory.

[0033] Furthermore, in step L4, the calculation of the distance between pedestrians and vehicles using the safe distance interaction algorithm includes:

[0034] L41. Based on the predicted pedestrian trajectory data, establish a safe distance interaction function J between pedestrians and vehicles.

[0035] ,

[0036] Among them, (c xk ,c yk ,c zk Let ) be the k-th trajectory point of the vehicle, and (r) xk ,r yk ,r zk Let be the k-th predicted trajectory point of the pedestrian, where N is a positive integer and N is the total number of samples.

[0037] L42. Based on the pedestrian and vehicle safe distance interaction function J, obtain the safe interaction distance data information between pedestrians and vehicles;

[0038] In step L5, adjusting the vehicle's interactive volume using an interactive volume and distance control algorithm includes:

[0039] L51. Based on the pedestrian and vehicle safety interaction distance data, establish an interaction volume and distance control function M.

[0040] M=(|P×B|)·J,

[0041] Where P is the volume adjustment matrix, B is the distance control matrix, and J is the pedestrian-vehicle safe distance interaction function;

[0042] L52. Based on the interactive volume and distance control function M, obtain the interactive volume data information between pedestrians and vehicles.

[0043] To achieve the above and other related objectives, the present invention also provides a human-machine interaction system based on an autonomous vehicle, including a computer device programmed or configured to perform the steps of any of the human-machine interaction methods based on an autonomous vehicle as described above.

[0044] To achieve the above and other related objectives, the present invention also provides a computer-readable storage medium storing a computer program programmed or configured to perform any of the human-machine interaction methods based on autonomous vehicles as described above.

[0045] The present invention has the following positive effects:

[0046] 1. This invention constructs a spatiotemporal map of roads and extracts pedestrian features to obtain pedestrian feature spatiotemporal map data information. It then uses a pedestrian relationship kernel function trajectory algorithm to calculate the pedestrian trajectory, solving the problem of inaccurate pedestrian trajectory prediction due to the lack of factors affecting pedestrian relationships, thereby improving the accuracy of human-computer interaction.

[0047] 2. This invention extracts features from a trained spatiotemporal graph convolutional neural network model to obtain feature data information of pedestrian trajectories. This feature data information of pedestrian trajectories is then input into a trained temporal extrapolator convolutional neural network model to predict pedestrian trajectories. This not only obtains predicted pedestrian trajectory data information but also makes the prediction of pedestrian trajectories more accurate. Furthermore, the human-computer interaction method is more reasonable and does not generate noise pollution, thereby improving the operating efficiency of autonomous vehicles.

[0048] 3. This invention uses a safe distance interaction algorithm to calculate the distance between pedestrians and vehicles, and combines this with an interaction volume and distance control algorithm to adjust the interaction volume of vehicles. This not only reduces noise pollution, but also ensures pedestrian safety and makes human-computer interaction smoother. Attached Figure Description

[0049] Figure 1 This is a schematic diagram of the method flow of the present invention;

[0050] Figure 2 This invention relates to a pedestrian relationship kernel function trajectory algorithm. Detailed Implementation

[0051] The exemplary embodiments of this disclosure are described below with reference to the accompanying drawings, including various details of the embodiments to aid understanding, and should be considered merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this disclosure. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0052] Example 1: As Figure 1 As shown, a human-machine interaction method based on autonomous vehicles includes:

[0053] L1. Autonomous vehicles drive on the road, acquiring real-time image data of the road based on onboard cameras and real-time point cloud data of the road based on onboard LiDAR, and preprocessing the data to obtain processed image and point cloud data of the road.

[0054] L2. Based on the processed road image and point cloud data, construct a spatiotemporal map of the road, extract pedestrian features to obtain pedestrian feature spatiotemporal map data, use the pedestrian relationship kernel function trajectory algorithm to calculate the pedestrian trajectory, and output the pedestrian trajectory data.

[0055] L3. Input the pedestrian trajectory data information into the trained spatiotemporal graph convolutional neural network model for feature extraction to obtain the feature data information of the pedestrian trajectory. Then, input the feature data information of the pedestrian trajectory into the trained temporal extrapolator convolutional neural network model to predict the pedestrian trajectory and output the predicted pedestrian trajectory data information.

[0056] L4. Based on the predicted pedestrian trajectory data, a safe distance interaction algorithm is used to calculate the distance between pedestrians and vehicles, and obtain safe interaction distance data between pedestrians and vehicles;

[0057] L5. Based on the safe interaction distance data between pedestrians and vehicles, the interaction volume of the vehicle is adjusted using an interaction volume and distance control algorithm, and the interaction volume data of pedestrians and vehicles is output.

[0058] In this embodiment, as Figure 2 As shown, in step L2, the calculation of the pedestrian trajectory using the pedestrian relationship kernel function trajectory algorithm includes:

[0059] L21. Based on the characteristic spatiotemporal graph data information of the pedestrians, characterize the relationship between each pedestrian and output the relationship characterization data information of the pedestrians;

[0060] L22. Based on the pedestrian relationship representation data, establish a pedestrian relationship kernel function Q.

[0061] ,

[0062] i≠j,

[0063] Where α is a constant parameter matrix, β is an inner product scaling matrix, and x i Let x be the i-th pedestrian relationship matrix. j Let ω be the j-th pedestrian relationship matrix, and let ω be the pedestrian simulation control parameter, to obtain pedestrian relationship data information;

[0064] L23. Based on the pedestrian relationship data, a cubic spline function is used to fit the pedestrian trajectory to obtain the pedestrian trajectory data.

[0065] In this embodiment, step L23, fitting the pedestrian's trajectory using a cubic spline function, includes:

[0066] L231. Based on the pedestrian relationship data, establish a cubic spline fitting function F for the pedestrian trajectory.

[0067] F=(Q×λ3)x 3 +(Q×λ2)x 2 +(Q×λ1)x+(Q×λ0),

[0068] Where Q represents pedestrian relationship data, λ0, λ1, λ2 and λ3 are the coefficient matrices of the cubic spline function, and x is the independent variable;

[0069] L232. Based on the cubic spline fitting function F of the pedestrian trajectory, the pedestrian trajectory is calculated to obtain the pedestrian trajectory data information.

[0070] Example 2: Based on the human-computer interaction method for autonomous vehicles in Example 1, the present invention will be further explained and described below.

[0071] like Figure 1 As shown, a human-machine interaction method based on autonomous vehicles includes:

[0072] L1. Autonomous vehicles drive on the road, acquiring real-time image data of the road based on onboard cameras and real-time point cloud data of the road based on onboard LiDAR, and preprocessing the data to obtain processed image and point cloud data of the road.

[0073] L2. Based on the processed road image and point cloud data, construct a spatiotemporal map of the road, extract pedestrian features to obtain pedestrian feature spatiotemporal map data, use the pedestrian relationship kernel function trajectory algorithm to calculate the pedestrian trajectory, and output the pedestrian trajectory data.

[0074] L3. Input the pedestrian trajectory data information into the trained spatiotemporal graph convolutional neural network model for feature extraction to obtain the feature data information of the pedestrian trajectory. Then, input the feature data information of the pedestrian trajectory into the trained temporal extrapolator convolutional neural network model to predict the pedestrian trajectory and output the predicted pedestrian trajectory data information.

[0075] L4. Based on the predicted pedestrian trajectory data, a safe distance interaction algorithm is used to calculate the distance between pedestrians and vehicles, and obtain safe interaction distance data between pedestrians and vehicles;

[0076] L5. Based on the safe interaction distance data between pedestrians and vehicles, the interaction volume of the vehicle is adjusted using an interaction volume and distance control algorithm, and the interaction volume data of pedestrians and vehicles is output.

[0077] In this embodiment, in step L3, the trained spatiotemporal graph convolutional neural network model is obtained by dividing the pedestrian trajectory data information into a training dataset and a test training dataset, and inputting the training dataset into the spatiotemporal graph convolutional neural network model for training and learning, determining the network parameters, and obtaining the trained spatiotemporal graph convolutional neural network model.

[0078] In this embodiment, the spatiotemporal graph convolutional neural network model includes a spatiotemporal graph convolutional input layer, a spatiotemporal graph convolutional kernel function, and a spatiotemporal graph convolutional output layer, wherein the spatiotemporal graph convolutional kernel function is G.

[0079] ,

[0080] Where A is the weight matrix, (x,y,z) are the pedestrian trajectory coordinates, and η0, η1 and η2 are the learning factor parameters of the corresponding pedestrian trajectory coordinates.

[0081] In this embodiment, the trained temporal extrapolator convolutional neural network model is obtained by dividing the feature data information of the pedestrian's trajectory into a training dataset and a test training dataset, and inputting the training dataset into the temporal extrapolator convolutional neural network model for training and learning, determining the network parameters, and obtaining the trained temporal extrapolator convolutional neural network model.

[0082] In this embodiment, the temporal extrapolator convolutional neural network model includes a temporal extrapolator input layer, a temporal extrapolator kernel function, and a temporal extrapolator kernel function H.

[0083] ,

[0084] ,

[0085] Where f is the observation function, X i Let X be the feature data information of the i-th pedestrian's trajectory, and O be the feature data information of the pedestrian's trajectory. i Let X be the Lagrange basis function for the i-th pedestrian. j Let n be the feature data information of the trajectory of the j-th pedestrian, and n be the total number of samples of the feature data of the pedestrian's trajectory.

[0086] In this embodiment, step L4, calculating the distance between pedestrians and vehicles using a safe distance interaction algorithm, includes:

[0087] L41. Based on the predicted pedestrian trajectory data, establish a safe distance interaction function J between pedestrians and vehicles.

[0088] ,

[0089] Among them, (cxk ,c yk ,c zk Let ) be the k-th trajectory point of the vehicle, and (r) xk ,r yk ,r zk Let be the k-th predicted trajectory point of the pedestrian, where N is a positive integer and N is the total number of samples.

[0090] L42. Based on the pedestrian and vehicle safe distance interaction function J, obtain the safe interaction distance data information between pedestrians and vehicles;

[0091] In step L5, adjusting the vehicle's interactive volume using an interactive volume and distance control algorithm includes:

[0092] L51. Based on the pedestrian and vehicle safety interaction distance data, establish an interaction volume and distance control function M.

[0093] M=(|P×B|)·J,

[0094] Where P is the volume adjustment matrix, B is the distance control matrix, and J is the pedestrian-vehicle safe distance interaction function;

[0095] L52. Based on the interactive volume and distance control function M, obtain the interactive volume data information between pedestrians and vehicles.

[0096] The present invention provides a human-machine interaction system based on an autonomous vehicle, including a computer device that is programmed or configured to perform the steps of any of the human-machine interaction methods based on an autonomous vehicle as described above.

[0097] The present invention provides a computer-readable storage medium storing a computer program that is programmed or configured to perform any of the human-machine interaction methods based on autonomous vehicles as described above.

[0098] Any references to memory, storage, database, or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0099] In summary, this invention not only provides more accurate prediction of pedestrian trajectories, but also offers a more rational human-computer interaction method and avoids noise pollution, thereby improving the operating efficiency of autonomous vehicles.

[0100] The specific embodiments described above do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A human-computer interaction method based on autonomous vehicles, characterized in that, The method includes: L1. Autonomous vehicles drive on the road, acquiring real-time image data of the road based on onboard cameras and real-time point cloud data of the road based on onboard LiDAR, and preprocessing the data to obtain processed image and point cloud data of the road. L2. Based on the processed road image and point cloud data, construct a spatiotemporal map of the road, extract pedestrian features to obtain pedestrian feature spatiotemporal map data, use the pedestrian relationship kernel function trajectory algorithm to calculate the pedestrian trajectory, and output the pedestrian trajectory data. L3. Input the pedestrian trajectory data information into the trained spatiotemporal graph convolutional neural network model for feature extraction to obtain the feature data information of the pedestrian trajectory. Then, input the feature data information of the pedestrian trajectory into the trained temporal extrapolator convolutional neural network model to predict the pedestrian trajectory and output the predicted pedestrian trajectory data information. L4. Based on the predicted pedestrian trajectory data, a safe distance interaction algorithm is used to calculate the distance between pedestrians and vehicles, and obtain safe interaction distance data between pedestrians and vehicles; L5. Based on the safe interaction distance data between pedestrians and vehicles, the interaction volume of the vehicle is adjusted using an interaction volume and distance control algorithm, and the interaction volume data of pedestrians and vehicles is output.

2. The human-computer interaction method based on autonomous vehicles according to claim 1, characterized in that, In step L2, the calculation of the pedestrian trajectory using the pedestrian relationship kernel function trajectory algorithm includes: L21. Based on the characteristic spatiotemporal graph data information of the pedestrians, characterize the relationship between each pedestrian and output the relationship characterization data information of the pedestrians; L22. Based on the pedestrian relationship representation data, establish a pedestrian relationship kernel function Q. ， i≠j, Where α is a constant parameter matrix, β is an inner product scaling matrix, and x i Let x be the i-th pedestrian relationship matrix. j Let ω be the j-th pedestrian relationship matrix, and let ω be the pedestrian simulation control parameter, to obtain pedestrian relationship data information; L23. Based on the pedestrian relationship data, a cubic spline function is used to fit the pedestrian trajectory to obtain the pedestrian trajectory data.

3. The human-computer interaction method based on autonomous vehicles according to claim 2, characterized in that, In step L23, fitting the pedestrian's trajectory using a cubic spline function includes: L231. Based on the pedestrian relationship data, establish a cubic spline fitting function F for the pedestrian trajectory. F=(Q×λ3)x 3 +(Q×λ2)x 2 +(Q×λ1)x+(Q×λ0), Where Q represents pedestrian relationship data, λ0, λ1, λ2 and λ3 are the coefficient matrices of the cubic spline function, and x is the independent variable; L232. Based on the cubic spline fitting function F of the pedestrian trajectory, the pedestrian trajectory is calculated to obtain the pedestrian trajectory data information.

4. The human-computer interaction method based on autonomous vehicles according to claim 1, characterized in that, In step L3, the trained spatiotemporal graph convolutional neural network model is obtained by dividing the pedestrian trajectory data information into a training dataset and a test training dataset, and inputting the training dataset into the spatiotemporal graph convolutional neural network model for training and learning, determining the network parameters, and obtaining the trained spatiotemporal graph convolutional neural network model.

5. The human-computer interaction method based on autonomous vehicles according to claim 4, characterized in that: The spatiotemporal graph convolutional neural network model includes a spatiotemporal graph convolutional input layer, a spatiotemporal graph convolutional kernel function, and a spatiotemporal graph convolutional output layer, wherein the spatiotemporal graph convolutional kernel function is G. ， Where A is the weight matrix, (x,y,z) are the pedestrian trajectory coordinates, and η0, η1 and η2 are the learning factor parameters of the corresponding pedestrian trajectory coordinates.

6. The human-computer interaction method based on autonomous vehicles according to claim 1, characterized in that: The trained temporal extrapolator convolutional neural network model is obtained by dividing the feature data information of the pedestrian's trajectory into a training dataset and a test training dataset, and inputting the training dataset into the temporal extrapolator convolutional neural network model for training and learning, determining the network parameters, and obtaining the trained temporal extrapolator convolutional neural network model.

7. The human-computer interaction method based on autonomous vehicles according to claim 6, characterized in that: The time extrapolator convolutional neural network model includes a time extrapolator input layer, a time extrapolator kernel function, and a time extrapolator kernel function H. ， ， Where f is the observation function, X i Let X be the feature data information of the i-th pedestrian's trajectory, and O be the feature data information of the pedestrian's trajectory. i Let X be the Lagrange basis function for the i-th pedestrian. j Let n be the feature data information of the trajectory of the j-th pedestrian, and n be the total number of samples of the feature data of the pedestrian's trajectory.

8. The human-computer interaction method based on autonomous vehicles according to claim 1, characterized in that, In step L4, the calculation of the distance between pedestrians and vehicles using the safe distance interaction algorithm includes: L41. Based on the predicted pedestrian trajectory data, establish a safe distance interaction function J between pedestrians and vehicles. ， Among them, (c xk ,c yk ,c zk Let (r) be the k-th trajectory point of the vehicle. xk ,r yk ,r zk Let be the k-th predicted trajectory point of the pedestrian, where N is a positive integer and N is the total number of samples. L42. Based on the pedestrian and vehicle safe distance interaction function J, obtain the safe interaction distance data information between pedestrians and vehicles; In step L5, adjusting the vehicle's interactive volume using an interactive volume and distance control algorithm includes: L51. Based on the pedestrian and vehicle safety interaction distance data, establish an interaction volume and distance control function M. M=(|P×B|)·J, Where P is the volume adjustment matrix, B is the distance control matrix, and J is the pedestrian-vehicle safe distance interaction function; L52. Based on the interactive volume and distance control function M, obtain the interactive volume data information between pedestrians and vehicles.

9. A human-machine interaction system based on autonomous vehicles, comprising computer equipment, characterized in that, The computer device is programmed or configured to perform the steps of the human-computer interaction method based on an autonomous vehicle as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that is programmed or configured to perform the human-computer interaction method based on an autonomous vehicle as described in any one of claims 1 to 8.

Images