Passenger autonomous mobile vehicle contract reaching method and system

Abstract

The invention discloses a contract reaching method and system for riding an autonomous mobile vehicle. According to the method, various modal information such as human faces, voiceprints and gestures is fused, a contract reaching method for riding an autonomous mobile vehicle in a public environment is provided, and a riding contract framework of the manned service robot is formed. The multi-mode biological feature fusion technology can integrate various biological feature input modes including voice, voiceprint, gesture and the like, and more flexible and safer identity authentication selection is provided for the user. And the user can select the most suitable identity authentication mode according to own preferences and habits, so that the reliability, rapidity and safety of information acquisition are further improved.

Description

Technical Field

[0001] The present application relates to the field of unmanned driving services, and more specifically, to a method and system for concluding a contract for a passenger autonomous mobile vehicle. Background Art

[0002] With the rapid development of digitalization and information technology, contractual agreements are no longer limited to traditional paper documents and face-to-face communication, but are gradually transitioning to digital and multimodal processes. Multimodal fusion, a key technology in this transformation, integrates information modalities from different sources and forms, providing reliability, speed, and security for contractual agreements while also ensuring that information access is non-exclusive.

[0003] During the contract negotiation process, a multimodal fusion system integrates multiple information sources and mutually verifies their authenticity and integrity. Authenticating identity through multiple biometric features significantly reduces the risk of fraud and forgery. Parallel processing and intelligent algorithm recognition technologies simultaneously process information from different modalities, rapidly extracting key information points and accelerating the recognition process. Multimodal fusion systems break the monolithic nature of traditional information acquisition channels and provide diverse information sources and methods. During the contract negotiation process, even if information from one modality cannot be obtained or verified for some reason, information from other modalities can still serve as an effective supplement, ensuring the smooth progress of the contract and improving the robustness of the system.

[0004] Although the application of single biometric technologies such as facial recognition and voiceprint recognition is relatively mature, they still have certain limitations. For example, facial recognition can be affected by factors such as lighting and obstructions, while voiceprint recognition can fail due to finger wear and sweating. Facial recognition technology has become the most widely used biometric technology due to its low cost and high recognition rate. However, at the same time, attacks against facial recognition systems are also emerging. For example, simple means such as using photos and videos can easily deceive the system. Therefore, it is necessary to propose a contract conclusion method and system for passenger autonomous vehicles to address the shortcomings of traditional recognition technologies, such as low reliability, speed, and security. Summary of the Invention

[0005] The content of this application is used to briefly introduce concepts that will be described in detail in the detailed description section below. The content of this application is not intended to identify key features or essential features of the technical solution for which protection is sought, nor is it intended to limit the scope of the technical solution for which protection is sought.

[0006] To address the technical issues mentioned in the background technology section above, some embodiments of the present application provide a method for fusing evidence for machine contract conclusion, including:

[0007] Receive identity feature information in multiple modalities;

[0008] Selecting identity feature information of any two modalities from multiple modalities of identity feature information;

[0009] Determine the target's identity information based on the identity feature information of the two selected modalities;

[0010] Determine whether the target's identity information corresponds to the pre-received identity information;

[0011] If the target's identity information does not correspond to the pre-received identity information, the identity feature information received in multiple modalities is returned, and information indicating that the identity authentication failed is fed back;

[0012] If the target's identity information matches the pre-received identity information, the various elements of information required to conclude the contract are collected and integrated;

[0013] Receive the positioning information of the front laser and determine the current position information;

[0014] Receive information input from the keyboard / touch screen and determine the destination;

[0015] Execute payment procedures to conclude the contract.

[0016] Furthermore, the identity feature information includes facial information, gesture information, voiceprint information, character information, ID card information and boarding pass information.

[0017] Furthermore, the facial information is preprocessed based on the RetinaFace algorithm;

[0018] Preprocess gesture information;

[0019] Preprocess the voiceprint information;

[0020] Obtain processed multi-modal information.

[0021] Further, receiving login information of the target;

[0022] Building a communication connection with the terminal used by the target based on the target's login information;

[0023] Receive the target's pre-stored identity information to complete the target's reservation.

[0024] Further, receiving a facial image of the target;

[0025] Face detection based on RetinaFace algorithm;

[0026] Use the Facenet network to encode and compare detected faces;

[0027] The facial recognition scores are calculated to obtain the facial image recognition comparison scores.

[0028] Furthermore, the feature vector of the facial image is compared with all feature vectors in the face database;

[0029] Calculate the Euclidean distance of the feature vector of each facial image in the face database;

[0030] Select the shortest Euclidean distance;

[0031] Determine whether the shortest Euclidean distance matches the threshold;

[0032] If the shortest Euclidean distance does not match the threshold, feedback is given as an unregistered prompt;

[0033] If the shortest Euclidean distance matches the threshold, the information is fed back to feed back the registration information.

[0034] Furthermore, we define the hypothesis space

[0035] Obtain the identified identities A, B, C and the empty set obtained from each modal fusion;

[0036] The probability distribution function (BPA) after weighted average preprocessing is used as the input of Dempster combination rule;

[0037] Calculate the BPA after fusion;

[0038] Compute the intersection and union of focal elements between different modes;

[0039] Update the fused BPA according to the rules of DS theory.

[0040] Furthermore, after defining the hypothesis space, the normalization constant formula is:

[0041]

[0042] Sum the joint mass functions of all hypothesis spaces whose intersections are not empty; for the hypothesis spaces whose intersections are not empty, calculate the joint mass function, and the synthesis rule is as follows:

[0043]

[0044] The fused BPA was obtained;

[0045] Use the trust function Bel(A) and the likelihood function Pl(A) to calculate the trust interval (Bel(A), Pl(A)) for each hypothesis;

[0046] Get the confidence interval for each combination;

[0047] Based on the confidence interval, the target identity is determined.

[0048] Furthermore, the identity features of the three modalities A, B, and C are integrated;

[0049] Obtain corresponding scores;

[0050] The three modal scores are fused to obtain the final decision result to determine whether the user's identity is authenticated.

[0051] Other embodiments of the present application provide an evidence fusion system for machine contract conclusion, including:

[0052] A processor, configured to execute the evidence fusion method for machine contract conclusion;

[0053] a front-view camera, communicatively connected to the processor;

[0054] A microphone, communicatively connected to the processor;

[0055] an identity card reader, communicatively connected to the processor;

[0056] a keyboard / touch screen, communicatively connected to the processor;

[0057] The front laser is communicatively connected with the processor.

[0058] The beneficial effect of this application is to provide a method and system for establishing a contract for a passenger autonomous mobile vehicle. This method integrates multimodal information, such as facial features, voiceprints, and gestures, to propose a method for establishing a contract for a passenger autonomous mobile robot in a public environment, forming a framework for establishing a contract for a passenger service robot. Multimodal biometric fusion technology can integrate multiple biometric input methods, including voice, voiceprints, and gestures, providing users with more flexible and secure identity authentication options. Users can choose the most suitable identity authentication method based on their preferences and habits, thereby further improving information security protection. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] The drawings constituting a part of this application are used to provide a further understanding of this application and make other features, purposes and advantages of this application more apparent. The drawings and descriptions of the exemplary embodiments of this application are used to explain this application and do not constitute an improper limitation on this application.

[0060] In addition, throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic and that the elements and components are not necessarily drawn to scale.

[0061] In the attached figure:

[0062] Figure 1 A flowchart of a method for integrating evidence for machine contract conclusion provided in one embodiment of the present application.

[0063] Figure 2 A system structure diagram of an evidence fusion system for achieving machine contracts provided in one embodiment of the present application.

[0064] Figure 3 An embodiment of the present application provides a method for integrating evidence for reaching a machine contract and a system for completing an overall flow chart of ride service interaction based on multimodality.

[0065] Figure 4 This is a system-wide data flow diagram of an evidence fusion system for achieving a machine contract provided in one embodiment of the present application.

[0066] Figure 5 A system structure diagram of an evidence fusion system for achieving machine contracts provided in one embodiment of the present application.

[0067] Reference numerals:

[0068] 100-processor; 200-front-view camera; 300-audio sensor; 400-keyboard / touch screen; 500-ID card reader; 600-front laser. DETAILED DESCRIPTION

[0069] Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be construed as being limited to the embodiments described herein. On the contrary, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of protection of the present disclosure.

[0070] It should also be noted that, for ease of description, only the parts related to the invention are shown in the drawings. In the absence of conflict, the embodiments and features in the embodiments of the present disclosure may be combined with each other.

[0071] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0072] It should be noted that the modifications of "one" and "multiple" mentioned in the present disclosure are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, they should be understood as "one or more".

[0073] The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with embodiments.

[0074] like Figure 1 As shown, the present application provides an evidence fusion method for machine contract conclusion.

[0075] S100, receiving identity feature information in multiple modalities.

[0076] S200: Select identity feature information of any two modalities from multiple modalities of identity feature information.

[0077] S300: Determine the target's identity information based on the identity feature information of the two selected modalities.

[0078] S400: Determine whether the target identity information corresponds to the pre-received identity information.

[0079] S500: If the identity information of the target does not correspond to the pre-received identity information, the identity feature information received in multiple modalities is returned, and information indicating that the identity authentication failed is fed back.

[0080] S600: If the target's identity information corresponds to the pre-received identity information, then collect and integrate the various elements of information required to conclude the contract.

[0081] S700: Execute the payment procedure to conclude the contract.

[0082] This method integrates multiple modalities, including facial, voiceprint, and gesture information, to propose a method for establishing implicit contracts for autonomous mobile robot rides in public environments, forming a framework for passenger service robot ride contracts. Multimodal biometric fusion technology can integrate multiple biometric input methods, including voice, voiceprint, and gesture, to provide users with more flexible and secure authentication options. Users can choose the most appropriate authentication method based on their preferences and habits, further enhancing information security. This addresses the inherent security limitations of traditional recognition technologies.

[0083] Reference Figures 2 to 4 As shown, in one embodiment of the present application, a method for integrating evidence for machine contract conclusion includes: identity feature information including multiple types of facial information, gesture information, voiceprint information, character information, ID card information and boarding pass information.

[0084] In the contract information collection part, it includes a front-facing camera, a microphone, an ID card reader, a front laser and a touch screen keyboard.

[0085] In the face acquisition module, the front vision sensor is responsible for acquiring input facial information. In the gesture acquisition module, the front vision sensor is responsible for acquiring input gesture information. In the voiceprint acquisition module, the audio sensor is responsible for acquiring input voiceprint and voice information. In the ID card acquisition module, the ID card reader acquires passenger ID information by scanning the ID card and boarding pass. In the touch screen acquisition module, the keyboard or touch screen is responsible for acquiring input text information. After acquiring modal information from multiple sensors, single-modal recognition and verification are performed.

[0086] This embodiment relates to a method for integrating evidence for machine contract conclusion.

[0087] Preprocess facial information based on the RetinaFace algorithm.

[0088] Based on preprocessing of gesture information.

[0089] Preprocess the voiceprint information.

[0090] Obtain processed multi-modal information.

[0091] Receive the target's login information.

[0092] Based on the target's login information, a communication connection is established with the terminal used by the target.

[0093] Receive the target's pre-stored identity information to complete the target's reservation.

[0094] Receive a target's facial image.

[0095] Perform face detection based on the RetinaFace algorithm.

[0096] The Facenet network is used to encode and compare the detected faces.

[0097] The facial recognition scores are calculated to obtain the facial image recognition comparison scores.

[0098] Compare the feature vector of the facial image with all feature vectors in the face database.

[0099] Calculate the Euclidean distance of the feature vector of each face image in the face database.

[0100] Select the shortest Euclidean distance.

[0101] Determine whether the shortest Euclidean distance matches the threshold.

[0102] If the shortest Euclidean distance does not match the threshold, a message is fed back indicating that the user has not registered.

[0103] If the shortest Euclidean distance matches the threshold, the information is fed back to feed back the registration information.

[0104] The face recognition module uses the RetinaFace algorithm to detect faces, and then uses the Facenet network to encode and compare the detected faces to obtain a face recognition score. The Retinaface algorithm uses Mobilenet0.25 as the backbone feature extraction network, integrates the Feature Pyramid Network (FPN) and Context Detection (SSH), detects faces of different scales, and improves the detection effect by enhancing the receptive field through environmental information. It can accurately locate facial features such as eyes, nose, mouth, etc. The algorithm shows superior performance in small faces and occlusion situations. The RetinaFace face detection algorithm is used to detect facial coordinates, and the coordinates are used to intercept and correct the face. Then, FaceNet is used to encode the corrected image to obtain a feature vector of length 128, which is a feature condensation of the input face and represents the identity of the input face. The feature vector of length 128 is compared with the feature vectors already existing in the database, and the Euclidean distance of each is calculated. After obtaining all Euclidean distances, the shortest Euclidean distance is selected and a check is performed to determine whether the Euclidean distance is less than a threshold. If this condition is met, it indicates that the input image has a corresponding image in the database and the user has registered. The voiceprint recognition module uses a hybrid network consisting of a convolutional neural network (CNN) and a long short-term memory network (LSTM). The CNN captures local features in the mid-frequency domain of the speech signal, while the LSTM captures the time series characteristics of the sound, enabling the model to effectively handle dynamically changing speech data. Before training the model, the audio file is preprocessed, including sampling rate adjustment, sound segmentation, and Mel-Frequency Cepstral Coefficient (MFCC) calculation. The raw audio data is converted into machine-interpretable feature vectors, which are then trained using a cross-entropy loss function, with accuracy used as the evaluation metric. The user first registers an audio file in the voiceprint database. For comparison, voiceprint features are obtained and the two audio files are compared for similarity. An audio file is then input and compared with the database for recognition. Administrators can remove registered users from the database. The feature data is obtained through the prediction function, and the diagonal cosine value is calculated using the feature data. The result can be used as their similarity, and the threshold of the similarity can be changed. According to the voiceprint sample at the time of user registration and the current voiceprint information, the Mel-frequency cepstral coefficient (MFCC) is used as the evaluation parameter of the voice feature similarity to calculate the voice recognition score and obtain the user voice recognition score. The gesture recognition module replaces the standard convolution layer of the C3 module in the YOLO-v5 algorithm with a switchable void convolution, and uses the improved YOLO-v5 algorithm to recognize and compare gestures to obtain the user gesture recognition score. The identity authentication part uses the integrated three-modal recognition network (face, gesture, voiceprint) to score separately, and fuses these scores to comprehensively judge the user's identity.

[0105] This embodiment relates to a method for integrating evidence for machine contract conclusion.

[0106] Defining the hypothesis space

[0107] Obtain the identified identities A, B, C and the empty set obtained by each modal fusion.

[0108] The probability distribution function (BPA) after weighted average preprocessing is used as the input of Dempster combination rule.

[0109] Calculate the BPA after fusion.

[0110] Compute the intersection and union of focal elements between different modes.

[0111] Update the fused BPA according to the rules of DS theory.

[0112] After defining the hypothesis space, the normalization constant formula is:

[0113]

[0114] Sum the joint mass functions of all hypothesis spaces whose intersections are not empty. For hypothesis spaces whose intersections are not empty, calculate the joint mass function, and the synthesis rule is as follows:

[0115]

[0116] The fused BPA was obtained.

[0117] The confidence interval (Bel(A), Pl(A)) of each hypothesis is calculated using the confidence function Bel(A) and the likelihood function Pl(A).

[0118] Get confidence intervals for each combination.

[0119] Based on the confidence interval, the target identity is determined.

[0120] Integrated identity features of the three modalities A, B, and C.

[0121] Get the corresponding score.

[0122] The three modal scores are fused to obtain the final decision result to determine whether the user's identity is authenticated.

[0123] The multimodal fusion authentication method fuses the score information obtained from the above three modalities through evidence theory, and derives a basic distribution probability distribution function based on the recognition results of each modality. This function represents the degree of certainty of the passenger's identity for each modality.

[0124] Determine the hypothesis space: The identified identities A, B, C and the empty set obtained by each sensor are obtained.

[0125] m1:m1(A)=0.8,m1(B)=0.2,m1(C)=0

[0126] m2:m2(A)=0.2,m2(B)=0.6,m2(C)=0

[0127] m3:m3(A)=0.2,m3(B)=0.6,m3(C)=0

[0128] The weighted average preprocessed BPA function is used as the input to the Dempster combination rule to calculate the fused BPA function. This process involves calculating the intersection and union of focal elements between different modes and updating the fused BPA function according to the rules of DS theory.

[0129] Sum the joint mass functions of all hypotheses whose intersections are not empty. For the hypotheses whose intersections are not empty, calculate the joint mass function. The synthesis rule is as follows:

[0130]

[0131] The following trust function Bel(A) and likelihood function Pl(A) are used to calculate the trust interval (Bel(A), Pl(A)) of each hypothesis, and the fusion recognition results within the trust interval prove that the passenger has passed the identity verification.

[0132]

[0133]

[0134] The contract conclusion method includes:

[0135] Regarding the contract conclusion part, after identity verification, the system will display the current location, estimated fare and travel time, etc. on the robot's screen for passengers to view.

[0136] After obtaining the passenger's destination, the system will draw a feasibility conclusion in the background, which means that the robot can deliver the passenger to the designated location within the specified time.

[0137] After the above contract information is collected, the robot will prompt the passenger to make the payment. After the passenger pays successfully, it means that the passenger agrees to enter into the contract. Otherwise, if the passenger refuses, the system will continue to prompt the passenger to make the payment.

[0138] After the passenger confirms that the information is correct, the robot completes the service, that is, confirms the service details and completes the ride contract.

[0139] like Figure 5As shown, a contract-reaching evidence fusion system includes:

[0140] The processor 100 is used to execute the evidence fusion method for achieving the machine contract.

[0141] The front-view camera 200 is in communication with the processor 100 .

[0142] The audio sensor 300 is communicatively connected to the processor 100 .

[0143] The keyboard / touch screen 400 is in communication with the processor 100 .

[0144] The ID card reader 500 is in communication with the processor 100 .

[0145] The front laser 500 is in communication with the processor 100 .

[0146] Example 1: After the robot is activated, it first checks the passenger's order information to determine if there are any special passenger transport tasks requiring special processing. If no such tasks exist, the robot enters its normal workflow and receives tasks from the dispatcher. During the preparation phase, the robot verifies the passenger's order information to ensure that the passenger is in the expected status. If the passenger's status does not meet expectations, the robot will move to a designated location to locate and assist the passenger. The robot then uploads all relevant order information to a cloud platform, enabling real-time updates and sharing of this information. To accurately locate the passenger, the robot uses identity authentication technology to ensure quick and accurate locating. Once located, the robot detects biometric information such as the passenger's face, gestures, and voiceprint to verify the passenger's identity and ensure that the service is being provided to the intended person. If the current user's identity matches the scheduled passenger, the robot queries the database to obtain the passenger's detailed information and relevant flight information. The robot's screen displays the passenger's current location, estimated fare, and travel time. After obtaining the passenger's destination, the system generates a feasibility conclusion in the backend, indicating that the robot can transport the passenger to the designated location within the specified time. The robot will then prompt the passenger to pay. Successful payment indicates the passenger's agreement to the contract. If the passenger declines, the system will continue prompting the passenger to pay. Once the passenger confirms the information is correct, the robot completes the service, confirming the ride details and finalizing the contract.

[0147] The present invention provides a multi-modal identity authentication method, which is divided into a registration phase and an authentication phase. Figure 2 , Figure 2 This is an identity authentication flow chart, which includes the following steps:

[0148] In Example 2, during the registration phase, the system uses the biometric recognition component to obtain the user's biometric information during registration. The user's basic personal information and biometric information during registration are stored separately in a database. The user's basic personal information during registration includes, among other things, the user's registered ID number. The biometric information during registration includes, among others, the user's facial image and voiceprint sample. During the authentication phase, the system uses the biometric recognition system to compare the recognized information with the user's biometric information during registration. The integrated three-modal recognition network identifies the user's identity characteristics and assigns corresponding scores. The three modal scores are then fused to produce a final decision to determine whether the user's identity has been authenticated. The camera and microphone of the biometric recognition module are used to obtain the user's current facial image, current gesture image, and current voiceprint information, respectively. Based on the user's facial image during registration, face detection is performed using the RetinaFace algorithm. The detected faces are then encoded and compared using the Facenet network. A facial recognition score is calculated to obtain a facial image recognition comparison score. Based on the user's registered voiceprint sample and the current voiceprint information, a voice recognition score is calculated using Mel-frequency Cepstral Coefficients (MFCCs) as an evaluation parameter for voice feature similarity. A modified YOLO-v5 algorithm is used to recognize and compare gestures. Finally, a multimodal recognition system is used to authenticate the user's biometrics. The authentication result is derived by fusing the facial image recognition score, gesture recognition score, and voice recognition score. If authentication passes, a successful authentication message is displayed, allowing login. If authentication fails, an authentication failure message is displayed.

[0149] The multimodal authentication method flow chart is shown in Figure 3 , Figure 3 It is a flow chart of the system's multimodal authentication method.

[0150] In Example 3, users store their user information in a database. During identity verification, their personal information is retrieved from the database, and the database returns the user information to the user. Passengers provide basic personal information, and the database returns their flight information to the passenger. The airport enters passenger flight (gate) information into the database, and the database provides booking information to the airport. The airline enters flight information into the database, and the database returns the revenue information to the airline. The user is a registered passenger in the system, and the passenger is a passenger who has made a reservation in the system that day.

[0151] The above description is only an illustration of some preferred embodiments of the present disclosure and the technical principles used. Those skilled in the art should understand that the scope of the invention involved in the embodiments of the present disclosure is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but should also cover other technical solutions formed by any combination of the above-mentioned technical features or their equivalent features without departing from the above-mentioned inventive concept. For example, the above-mentioned features are replaced with (but not limited to) technical features with similar functions disclosed in the embodiments of the present disclosure.

Claims

1. A method for integrating evidence for reaching a contract for a passenger autonomous vehicle, characterized in that: include: Receive identity feature information in multiple modalities; Selecting identity feature information of any two modalities from multiple modalities of identity feature information; Determine the target's identity information based on the identity feature information of the two selected modalities; Determine whether the target's identity information corresponds to the pre-received identity information; If the target's identity information does not correspond to the pre-received identity information, the identity feature information received in multiple modalities is returned, and information indicating that the identity authentication failed is fed back; If the target's identity information matches the pre-received identity information, the various elements of information required to conclude the contract are collected and integrated; Receive the positioning information of the front laser and determine the current position information; Receive information input from the keyboard / touch screen and determine the destination; Execute payment procedures to conclude the contract.

2. The method for integrating evidence for reaching a contract for an autonomous passenger vehicle according to claim 1, characterized in that: The receiving of identity feature information in multiple modes includes: Identity feature information includes facial information, gesture information, voiceprint information, character information, ID card information and boarding pass information.

3. The method for integrating evidence for reaching a contract for an autonomous passenger vehicle according to claim 2, characterized in that: The receiving of identity feature information in multiple modalities further includes: Preprocess facial information based on the RetinaFace algorithm; Preprocess gesture information; Preprocess the voiceprint information; Obtain processed multi-modal information.

4. The method for integrating evidence for reaching a contract for an autonomous passenger vehicle according to claim 3, characterized in that: After receiving identity feature information of multiple modalities, the method includes: Receive the target's login information; Building a communication connection with the terminal used by the target based on the target's login information; Receive the target's pre-stored identity information to complete the target's reservation.

5. The method for integrating evidence for reaching a contract for an autonomous passenger vehicle according to claim 4, characterized in that: The determining of the target's identity information based on the identity feature information of the two selected modalities includes: Receive the target's facial image; Perform face detection based on the RetinaFace algorithm; Use the Facenet network to encode and compare detected faces; The facial recognition scores are calculated to obtain the facial image recognition comparison scores.

6. The method for integrating evidence for reaching a contract for an autonomous passenger vehicle according to claim 5, characterized in that: The determining of the target's identity information based on the identity feature information of the two selected modalities further includes: Compare the feature vector of the facial image with all feature vectors in the face database; Calculate the Euclidean distance of the feature vector of each facial image in the face database; Select the shortest Euclidean distance; Determine whether the shortest Euclidean distance matches the threshold; If the shortest Euclidean distance does not match the threshold, feedback is given as an unregistered prompt; If the shortest Euclidean distance matches the threshold, the information is fed back to feed back the registration information.

7. The method for integrating evidence for reaching a contract for an autonomous passenger vehicle according to claim 6, characterized in that: The determining of the target's identity information based on the identity feature information of the two selected modalities further includes: Defining the hypothesis space Obtain the identified identities A, B, C and the empty set obtained from each modal fusion; The probability distribution function (BPA) after weighted average preprocessing is used as the input of Dempster combination rule; Calculate the BPA after fusion; Compute the intersection and union of focal elements between different modes; Update the fused BPA according to the rules of DS theory.

8. The method for integrating evidence for reaching a contract for an autonomous passenger vehicle according to claim 7, characterized in that: The calculation of the intersection and union of focal elements between different modes includes: After defining the hypothesis space, the normalization constant formula is: Sum the joint mass functions of all hypothesis spaces whose intersections are not empty; for the hypothesis spaces whose intersections are not empty, calculate the joint mass function, and the synthesis rule is as follows: The fused BPA was obtained; Use the trust function Bel(A) and the likelihood function Pl(A) to calculate the trust interval (Bel(A), Pl(A)) for each hypothesis; Get the confidence interval for each combination; Based on the confidence interval, the target identity is determined.

9. The evidence fusion method for machine contract conclusion according to claim 8, characterized in that: The determining whether the target identity information corresponds to the pre-received identity information includes: Integrated identity features of the three modalities A, B, and C; Obtain corresponding scores; The three modal scores are fused to obtain the final decision result to determine whether the user's identity is authenticated.

10. An evidence fusion system for machine contract conclusion, characterized in that: include: A processor, configured to execute the evidence fusion method for reaching a mobile vehicle contract as claimed in any one of claims 1 to 9; a front-view camera, communicatively connected to the processor; A microphone, communicatively connected to the processor; an identity card reader, communicatively connected to the processor; a keyboard / touch screen, communicatively connected to the processor; The front laser is communicatively connected with the processor.

Images