Intelligent interactive charging system for two-wheeled electric vehicle

Abstract

The invention discloses an intelligent interactive charging system of a two-wheeled electric vehicle, belongs to the technical field of charging, and aims to solve the technical problem of how to realize intelligent and humanized intelligent interactive charging of the two-wheeled electric vehicle and improve user experience. A double-spectrum visual perception module, an annular LED matrix projection module, a face emotion recognition module, a multi-language voice interaction module, an identity verification module and an old people care module are arranged outside the charging pile; the double-spectrum visual perception module and the annular LED matrix projection module are used for jointly realizing a night visual guidance function; the face emotion recognition module adopts a camera and an edge computing chip and is used for locally analyzing facial expression features of a user in real time and acquiring an emotion analysis result; the multi-language voice interaction module adopts a microphone array to carry out noise reduction and sound source positioning.

Description

Technical Field

[0001] This invention relates to the field of charging technology, specifically to an intelligent interactive charging system for two-wheeled electric vehicles. Background Technology

[0002] With the increasing popularity of two-wheeled electric vehicles, the demand for public charging stations is growing. However, existing charging stations are functionally limited and lack user-friendly interfaces, presenting the following pain points:

[0003] (1) It is difficult to quickly locate available charging stations at night, resulting in a poor user experience and potential safety hazards;

[0004] (2) It is unable to perceive user emotions and may exacerbate user anxiety when queuing or experiencing operational difficulties.

[0005] (3) The interactive language is limited and cannot meet the needs of elderly people in dialect areas or foreigners.

[0006] (4) The charging power allocation is rigid and cannot be optimized according to the user's urgency.

[0007] (5) It is not user-friendly for elderly users, the operation is complicated and there is a lack of effective channels for help.

[0008] While some charging stations in the current technology have introduced simple voice prompts or screen interactions, they have failed to achieve the fusion perception and intelligent decision-making of multimodal information, and lack the ability to provide personalized services based on user emotions and identity.

[0009] Therefore, how to achieve intelligent and user-friendly smart interactive charging for two-wheeled electric vehicles and improve user experience is a technical problem that urgently needs to be solved. Summary of the Invention

[0010] The technical objective of this invention is to provide an intelligent interactive charging system for two-wheeled electric vehicles to address the problem of how to achieve intelligent and user-friendly intelligent interactive charging for two-wheeled electric vehicles and improve user experience.

[0011] The technical objective of this invention is achieved as follows: a two-wheeled electric vehicle intelligent interactive charging system, which includes a charging pile, a main control module installed inside the charging pile, and a dual-spectrum visual perception module, a ring LED matrix projection module, a facial emotion recognition module, a multi-language voice interaction module, an identity verification module, and an elderly care module installed outside the charging pile.

[0012] The dual-spectrum visual perception module and the ring-shaped LED matrix projection module are used together to achieve nighttime visual guidance function; among them, the ring-shaped LED matrix projection module projects a high-brightness, high-recognition light spot according to the command to achieve long-distance guidance;

[0013] The facial emotion recognition module uses a camera and an edge computing chip to analyze users' facial expression features (such as the corners of the mouth, eyebrows, and wrinkles around the eyes) locally in real time and obtain emotion analysis results, protecting user privacy.

[0014] The multilingual voice interaction module uses a microphone array for noise reduction and sound source localization to ensure accurate voice recognition. Through a hybrid "cloud + local" architecture, it supports rich dialect recognition and can also ensure basic English interaction when there is no network.

[0015] The identity verification module integrates an ID card reader, which connects to the main control module via a USB interface for accurate identity binding and credit verification, and is suitable for elderly care scenarios;

[0016] The elderly care module uses an independent physical "one-click help" button and a voice amplification circuit to ensure that the most direct and reliable help is provided in critical moments.

[0017] As a preferred option, the main control module adopts an embedded SoC (system-on-a-chip) based on the ARM architecture. The main control module uses NVIDIA Jetson Nano or TX2 series, or Rockchip RK3588 high-performance processor, which has powerful general computing capabilities and a dedicated AI processing unit (NPU).

[0018] The main control module runs a customized Linux operating system (such as Ubuntu 18.04LTS), on which Docker containers are deployed to manage the applications of each functional module.

[0019] As a preferred option, the dual-spectrum visual perception module uses a dual-spectrum network camera with visible light and thermal imaging to accurately determine the brightness of the environment and to perform real-time analysis of the video stream using the OpenCV computer vision library to detect whether the parking space is occupied by a vehicle.

[0020] The dual-spectrum visual perception module is installed on the top of the charging pile and fixed in a way that can be adjusted in pitch angle. The shooting area of ​​the dual-spectrum visual perception module covers the area in front of the charging pile and the adjacent parking spaces.

[0021] As a preferred embodiment, the ring-shaped LED matrix projection module consists of a ring array of multiple LED beads (such as OSRAM or Cree brand) with a brightness of over 100 lumens, and is equipped with a Fresnel lens or microlens array on the outside for focusing and shaping the light spot; the ring-shaped LED matrix projection module is covered with a high-strength PC or tempered glass panel with a protection level of IP65, which is dustproof and waterproof.

[0022] The main control module controls an LED driver chip via the I2C or SPI protocol. The LED driver chip performs independent PWM (pulse width modulation) control on each LED, thereby dynamically generating different patterns and text, such as an "empty pile" arrow, a charging progress bar, or a "fault" cross icon.

[0023] As a preferred option, the facial emotion recognition module uses a wide-angle camera with a near-infrared fill light (such as the Sony IMX series sensor) to ensure that clear facial images can be captured even at night.

[0024] A lightweight deep learning model (such as a modified MobileNet or SqueezeNet) is deployed on the NPU of the main control module for emotion recognition. The lightweight deep learning model is trained on the FER-2013 or AffectNet public datasets to identify basic emotions such as calm, joy, anxiety, anger, and surprise, and output discrete emotion labels and a continuous emotion confidence score (0-1). For example, the system only determines a valid negative emotion event when it is identified as "anxiety" and the confidence score is higher than 0.7.

[0025] As a preferred option, the multilingual voice interaction module includes a ring microphone array (6 or 8 microphones) and a speaker. The ring microphone array is used to achieve sound source localization and environmental noise reduction; the speaker uses a full-range, waterproof unit to ensure clear sound quality.

[0026] The multilingual voice interaction module includes online and offline modes;

[0027] The online mode specifically involves: integrating the SDK of the voice open platform; when the network connection is good, the user's voice data is encrypted and uploaded to the cloud, where it is recognized and semantically understood using a cloud-based dialect recognition engine (such as supporting Cantonese, Sichuanese, Shanghainese, etc.);

[0028] The offline mode is as follows: a compact but efficient English speech recognition and synthesis engine (such as one based on the PocketSphinx or Vosk library) is pre-installed in the local storage space; when a network interruption is detected or multiple cloud recognition failures are detected, the system automatically switches to the offline English mode to ensure basic interactions for foreign users (such as "Start charging", "Stop", "Help").

[0029] Even better, the independent physical "one-click help" button of the elderly care module adopts a physical anti-accidental touch emergency button with LED backlight. The emergency button has an IP67 protection rating and contains a normally open switch signal connected to the GPIO port of the main control module.

[0030] As a preferred option, the working process of this system is as follows:

[0031] S1. Environmental Perception and Automatic Guidance: The dual-spectrum visual perception module continuously captures environmental images, and the main control module calculates the average brightness value of the environmental images. When the average brightness value of the environmental images is lower than a preset threshold (e.g., 10 lux, indicating dusk or late night), it is determined to be a "dark environment." Once it is determined that it is in a dark environment and its own status is idle, the main control module immediately sends a command to the ring LED matrix projection module to control the ring LED matrix projection module to project a clear green "empty pile" arrow icon onto the ground below. At the same time, the video stream is analyzed in real time using background subtraction or YOLO object detection algorithms to determine whether the charging pile and adjacent parking spaces are occupied by vehicles. If the parking space is occupied, the ring LED matrix projection module projects a red "occupied" icon or no light.

[0032] S2. User Approach and Emotion Perception: When the dual-spectrum visual perception module detects a heat source (i.e., a person) entering the charging area (e.g., within 5 meters), the facial emotion recognition thread is triggered, the facial emotion recognition module is activated, and the user's facial image is captured. The user's facial image is sent to the emotion recognition model on the local NPU for analysis. The control module quantifies and classifies the user's emotion based on the emotion label and confidence level output by the lightweight deep learning model, specifically: Level 0 (calm / pleasant): normal interaction mode; Level 1 (mild anxiety): mark the user and prepare to initiate gentle soothing; Level 2 (severe anxiety / anger): immediately trigger advanced soothing mode and notify the power scheduling system.

[0033] S3. Personalized Voice Interaction and Reassurance: A welcome message, such as "Hello, welcome to the smart charging station," is played through the speaker of the multilingual voice interaction module. Simultaneously, the circular microphone array of the multilingual voice interaction module begins listening. If the emotion level is Level 1, reassuring words are incorporated into the standard operation instructions, such as "Don't worry, the operation is very simple; I will guide you step by step." If the emotion level is Level 2, a proactive inquiry is made: "You seem a little anxious; would you like me to prioritize your needs?" A short, soothing background music may also be played. After the user replies, cloud-based dialect recognition is attempted first. If cloud recognition is successful, the same dialect is used in the reply. If the user speaks English or the cloud connection times out, the system switches to the local English speech recognition and synthesis engine.

[0034] S4. Identity Verification and Credit Check: The voice prompts the user to verify their identity: "Please swipe your card or scan the QR code." If the facial emotion recognition module initially determines that the user is over 65 years old based on facial feature analysis (such as wrinkle depth and facial contour), a special prompt will be given: "We have detected that you may be an elderly user. We recommend using your ID card for verification to obtain exclusive services." After the user swipes their card or scans the code, the system connects to the backend user database via 4G / 5G or Ethernet to obtain the user's historical credit score. The user's historical credit score is calculated based on past behavior such as timely payment and proper parking.

[0035] S5. Dynamic Power Scheduling: The main control module uses the user's mood level and historical credit score as two key weight parameters, encrypts them, and reports them to the cloud-based charging dispatch management center. The cloud-based charging dispatch management center makes decisions based on a combination of user weights and grid constraints. It sorts all online charging requests according to their weight scores and, within the total power limit, prioritizes allocating higher power to users with higher weight scores. For example, the default power is 500W, but for high-weight users, it can be temporarily increased to 800W to shorten their charging time. User weight score = A * mood level + B * credit score; A and B are adjustable coefficients, for example, A = 0.6, B = 0.4. Users with high anxiety levels and good credit have higher weight scores. Grid constraints refer to the current total grid load and requests from other charging piles on the same line.

[0036] S6. Charging Execution and Continuous Monitoring: After the user connects the charging gun and confirms, charging begins. The display screen or LED spot will show the charging progress. If the user is waiting, their emotional changes can be detected intermittently, and the interactive content can be dynamically adjusted. A dual-spectrum network camera continuously monitors the charging interface temperature (thermal imaging function) to prevent overheating failure. Once the "One-Click Help" button is pressed, the main control module immediately interrupts the current task, initiates a video call request with the remote customer service center via the 4G network, and pushes the real-time video stream of this charging pile and user information to the customer service screen.

[0037] The intelligent interactive charging system for two-wheeled electric vehicles of the present invention has the following advantages:

[0038] (i) This invention realizes intelligent interactive charging for two-wheeled electric vehicles with intelligent, humanized, and high user experience, and can understand users, respond proactively and provide emotional care.

[0039] (II) This invention proactively discovers user needs (nighttime charging station location) through visual perception and uses projection for non-intrusive guidance. It senses user emotions at the beginning of the interaction, incorporates the important but neglected dimension of "emotion" into the interaction strategy, provides warm service, and transforms users from anonymous individuals into entities with historical data and specific attributes (such as age) through identity recognition, thereby providing highly personalized services (such as power priority and voice amplification). It combines soft indicators such as emotion and credit with hard constraints of the power grid to achieve dynamic optimization of resource allocation and improve the overall system efficiency and fairness.

[0040] (iii) This invention can efficiently run complex computer vision and deep learning models (such as emotion recognition) on the local machine, reduce dependence on cloud networks, and ensure response speed and user privacy;

[0041] (iv) This invention greatly improves the user experience, especially in the dark or when anxious, by guiding the user with nighttime light spots and calming emotions through emotion recognition;

[0042] (V) This invention improves charging turnover rate and grid utilization efficiency by dynamically scheduling power based on emotion and credit.

[0043] (vi) This invention adopts multilingual / dialect support and special care for the elderly (two-factor verification, one-click help), breaks down usage barriers, demonstrates technological and humanistic care, improves the quality of public facility services, and enhances inclusiveness;

[0044] (vii) This invention integrates and analyzes multi-source data such as vision (environment, parking space, face), voice, user identity and real-time emotion through the main control system to achieve intelligent interaction and decision-making, rather than a simple superposition of functions;

[0045] (viii) This invention quantifies and classifies the user emotions identified by AI, and makes them the core control parameters for triggering the "soothing mode" and dynamically adjusting the charging power priority, so that the device has the ability to respond "emotionally".

[0046] (ix) This invention integrates a specific combination of modules such as dual-spectrum vision, LED matrix projection, emotion recognition, multilingual interaction, and identity verification, as well as their collaborative logical architecture. It consists of a complete process of environmental perception → emotion recognition → personalized interaction → dynamic power scheduling, and in particular, a mathematical model that uses emotion level and user credit as key input parameters for the power allocation algorithm. Attached Figure Description

[0047] The invention will be further described below with reference to the accompanying drawings.

[0048] Appendix Figure 1 A schematic diagram of the structure of an intelligent interactive charging system for two-wheeled electric vehicles. Detailed Implementation

[0049] The following detailed description of an intelligent interactive charging system for a two-wheeled electric vehicle according to the present invention is based on the accompanying drawings and specific embodiments.

[0050] Example:

[0051] As attached Figure 1 As shown, this embodiment provides an intelligent interactive charging system for two-wheeled electric vehicles. The system includes a charging pile, a main control module inside the charging pile, and a dual-spectrum visual perception module, a ring LED matrix projection module, a facial emotion recognition module, a multi-language voice interaction module, an identity verification module, and an elderly care module outside the charging pile.

[0052] The dual-spectrum visual perception module and the ring-shaped LED matrix projection module are used together to achieve nighttime visual guidance function; among them, the ring-shaped LED matrix projection module projects a high-brightness, high-recognition light spot according to the command to achieve long-distance guidance;

[0053] The facial emotion recognition module uses a camera and an edge computing chip to analyze users' facial expression features (such as the corners of the mouth, eyebrows, and wrinkles around the eyes) locally in real time and obtain emotion analysis results, protecting user privacy.

[0054] The multilingual voice interaction module uses a microphone array for noise reduction and sound source localization to ensure accurate voice recognition. Through a hybrid "cloud + local" architecture, it supports rich dialect recognition and can also ensure basic English interaction when there is no network.

[0055] The identity verification module integrates an ID card reader, which connects to the main control module via a USB interface for accurate identity binding and credit verification, and is suitable for elderly care scenarios;

[0056] The elderly care module uses an independent physical "one-click help" button and a voice amplification circuit to ensure that the most direct and reliable help is provided in critical moments.

[0057] In this embodiment, the main control module adopts an embedded SoC (System-on-a-Chip) based on the ARM architecture. The main control module uses NVIDIA Jetson Nano or TX2 series, or Rockchip RK3588 high-performance processor, which has powerful general computing capabilities and a dedicated AI processing unit (NPU). It can efficiently run complex computer vision and deep learning models (such as emotion recognition) on the local machine, reduce dependence on cloud networks, and ensure response speed and user privacy.

[0058] The main control module runs a customized Linux operating system (such as Ubuntu 18.04LTS), on which Docker containers are deployed to manage the applications of each functional module.

[0059] In this embodiment, the dual-spectrum visual perception module uses a dual-spectrum network camera with visible light and thermal imaging to accurately determine the ambient brightness and performs real-time analysis of the video stream using the OpenCV computer vision library to detect whether the parking space is occupied by a vehicle.

[0060] The dual-spectrum visual perception module is installed on the top of the charging pile and fixed in a way that can be adjusted in pitch angle. The shooting area of ​​the dual-spectrum visual perception module covers the area in front of the charging pile and the adjacent parking spaces.

[0061] For example, some models from Hikvision or Dahua Technology have thermal imaging modules that can be used to detect objects (such as people and vehicles) at night and in inclement weather, while visible light modules are used for high-definition image capture during the day.

[0062] In this embodiment, the ring-shaped LED matrix projection module consists of a ring array of multiple LED beads (such as OSRAM or Cree brand) with a brightness of over 100 lumens. It is equipped with a Fresnel lens or microlens array on the outside for focusing and shaping the light spot. The ring-shaped LED matrix projection module is covered with a high-strength PC or tempered glass panel with a protection level of IP65, which is dustproof and waterproof.

[0063] The main control module controls an LED driver chip via the I2C or SPI protocol. The LED driver chip performs independent PWM (pulse width modulation) control on each LED, thereby dynamically generating different patterns and text, such as an "empty pile" arrow, a charging progress bar, or a "fault" cross icon.

[0064] In this embodiment, the facial emotion recognition module uses a wide-angle camera with a near-infrared fill light (such as a Sony IMX series sensor) to ensure that clear facial images can be captured even at night.

[0065] A lightweight deep learning model (such as a modified MobileNet or SqueezeNet) is deployed on the NPU of the main control module for emotion recognition. The lightweight deep learning model is trained on the FER-2013 or AffectNet public datasets to identify basic emotions such as calm, joy, anxiety, anger, and surprise, and output discrete emotion labels and a continuous emotion confidence score (0-1). For example, the system only determines a valid negative emotion event when it is identified as "anxiety" and the confidence score is higher than 0.7.

[0066] The multilingual voice interaction module in this embodiment includes a ring microphone array (6 or 8 microphones) and a speaker. The ring microphone array is used to achieve sound source localization and environmental noise reduction; the speaker uses a full-frequency, waterproof unit to ensure clear sound quality.

[0067] The multilingual voice interaction module includes online and offline modes;

[0068] The online mode specifically involves: integrating the SDK of the voice open platform; when the network connection is good, the user's voice data is encrypted and uploaded to the cloud, where it is recognized and semantically understood using a cloud-based dialect recognition engine (such as supporting Cantonese, Sichuanese, Shanghainese, etc.);

[0069] The offline mode is as follows: a compact but efficient English speech recognition and synthesis engine (such as one based on the PocketSphinx or Vosk library) is pre-installed in the local storage space; when a network interruption is detected or multiple cloud recognition failures are detected, the system automatically switches to the offline English mode to ensure basic interactions for foreign users (such as "Start charging", "Stop", "Help").

[0070] In this embodiment, the independent physical "one-click emergency call" button of the elderly care module is a physical, backlit LED emergency button designed to prevent accidental touches. The emergency button has an IP67 protection rating and internally contains a normally open switch signal connected to the GPIO port of the main control module. After confirming that the user is an elderly person, the control module sends a command to the audio driver to increase the speaker's output gain by 30%-50% and enhance the mid-low frequencies, making the voice louder and easier to hear.

[0071] The working process of this system is as follows:

[0072] S1. Environmental Perception and Automatic Guidance: The dual-spectrum visual perception module continuously captures environmental images, and the main control module calculates the average brightness value of the environmental images. When the average brightness value of the environmental images is lower than a preset threshold (e.g., 10 lux, indicating dusk or late night), it is determined to be a "dark environment." Once it is determined that it is in a dark environment and its own status is idle, the main control module immediately sends a command to the ring LED matrix projection module to control the ring LED matrix projection module to project a clear green "empty pile" arrow icon onto the ground below. At the same time, the video stream is analyzed in real time using background subtraction or YOLO object detection algorithms to determine whether the charging pile and adjacent parking spaces are occupied by vehicles. If the parking space is occupied, the ring LED matrix projection module projects a red "occupied" icon or no light.

[0073] S2. User Approach and Emotion Perception: When the dual-spectrum visual perception module detects a heat source (i.e., a person) entering the charging area (e.g., within 5 meters), the facial emotion recognition thread is triggered, the facial emotion recognition module is activated, and the user's facial image is captured. The user's facial image is sent to the emotion recognition model on the local NPU for analysis. The control module quantifies and classifies the user's emotion based on the emotion label and confidence level output by the lightweight deep learning model, specifically: Level 0 (calm / pleasant): normal interaction mode; Level 1 (mild anxiety): mark the user and prepare to initiate gentle soothing; Level 2 (severe anxiety / anger): immediately trigger advanced soothing mode and notify the power scheduling system.

[0074] S3. Personalized Voice Interaction and Reassurance: A welcome message, such as "Hello, welcome to the smart charging station," is played through the speaker of the multilingual voice interaction module. Simultaneously, the circular microphone array of the multilingual voice interaction module begins listening. If the emotion level is Level 1, reassuring words are incorporated into the standard operation instructions, such as "Don't worry, the operation is very simple; I will guide you step by step." If the emotion level is Level 2, a proactive inquiry is made: "You seem a little anxious; would you like me to prioritize your needs?" A short, soothing background music may also be played. After the user replies, cloud-based dialect recognition is attempted first. If cloud recognition is successful, the same dialect is used in the reply. If the user speaks English or the cloud connection times out, the system switches to the local English speech recognition and synthesis engine.

[0075] S4. Identity Verification and Credit Check: The voice prompts the user to verify their identity: "Please swipe your card or scan the QR code." If the facial emotion recognition module initially determines that the user is over 65 years old based on facial feature analysis (such as wrinkle depth and facial contour), a special prompt will be given: "We have detected that you may be an elderly user. We recommend using your ID card for verification to obtain exclusive services." After the user swipes their card or scans the code, the system connects to the backend user database via 4G / 5G or Ethernet to obtain the user's historical credit score. The user's historical credit score is calculated based on past behavior such as timely payment and proper parking.

[0076] S5. Dynamic Power Scheduling: The main control module uses the user's mood level and historical credit score as two key weight parameters, encrypts them, and reports them to the cloud-based charging dispatch management center. The cloud-based charging dispatch management center makes decisions based on a combination of user weights and grid constraints. It sorts all online charging requests according to their weight scores and, within the total power limit, prioritizes allocating higher power to users with higher weight scores. For example, the default power is 500W, but for high-weight users, it can be temporarily increased to 800W to shorten their charging time. User weight score = A * mood level + B * credit score; A and B are adjustable coefficients, for example, A = 0.6, B = 0.4. Users with high anxiety levels and good credit have higher weight scores. Grid constraints refer to the current total grid load and requests from other charging piles on the same line.

[0077] S6. Charging Execution and Continuous Monitoring: After the user connects the charging gun and confirms, charging begins. The display screen or LED spot will show the charging progress. If the user is waiting, their emotional changes can be detected intermittently, and the interactive content can be dynamically adjusted. A dual-spectrum network camera continuously monitors the charging interface temperature (thermal imaging function) to prevent overheating failure. Once the "One-Click Help" button is pressed, the main control module immediately interrupts the current task, initiates a video call request with the remote customer service center via the 4G network, and pushes the real-time video stream of this charging pile and user information to the customer service screen.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A two-wheeled electric vehicle intelligent interactive charging system, characterized in that, The system includes a charging pile, which contains a main control module. Externally, the charging pile is equipped with a dual-spectrum visual perception module, a ring-shaped LED matrix projection module, a facial emotion recognition module, a multi-language voice interaction module, an identity verification module, and an elderly care module. The dual-spectrum visual perception module and the ring-shaped LED matrix projection module are used together to achieve nighttime visual guidance function; among them, the ring-shaped LED matrix projection module projects a high-brightness, high-recognition light spot according to the command to achieve long-distance guidance; The facial emotion recognition module uses a camera and an edge computing chip to analyze users' facial expression features locally in real time and obtain emotion analysis results, thus protecting user privacy. The multilingual voice interaction module uses a microphone array for noise reduction and sound source localization to ensure accurate voice recognition. Through a hybrid "cloud + local" architecture, it supports rich dialect recognition and can also ensure basic English interaction when there is no network. The identity verification module integrates an ID card reader, which connects to the main control module via a USB interface for accurate identity binding and credit verification, and is suitable for elderly care scenarios; The elderly care module uses an independent physical "one-click help" button and a voice amplification circuit to ensure that the most direct and reliable help is provided in critical moments.

2. The two-wheeled electric vehicle intelligent interactive charging system of claim 1, wherein, The main control module adopts an ARM-based embedded SoC, which uses NVIDIA Jetson Nano or TX2 series, or Rockchip RK3588 high-performance processor, and has powerful general computing capabilities and a dedicated AI processing unit. The main control module runs a customized Linux operating system, on which Docker containers are deployed to manage the applications of each functional module.

3. The two-wheeled electric vehicle intelligent interactive charging system of claim 1, wherein, The dual-spectrum visual perception module uses a dual-spectrum network camera with visible light and thermal imaging to accurately determine the brightness of the environment and uses the OpenCV computer vision library to analyze the video stream in real time to detect whether the parking space is occupied by a vehicle. The dual-spectrum visual perception module is installed on the top of the charging pile and fixed in a way that can be adjusted in pitch angle. The shooting area of ​​the dual-spectrum visual perception module covers the area in front of the charging pile and the adjacent parking spaces.

4. The two-wheeled electric vehicle intelligent interactive charging system of claim 1, wherein, The ring-shaped LED matrix projection module consists of a ring array of multiple lumens LED beads, and is equipped with a Fresnel lens or microlens array on the outside for focusing and shaping the light spot; the ring-shaped LED matrix projection module is covered with a high-strength PC or tempered glass panel, with a protection level of IP65, making it dustproof and waterproof; The main control module controls an LED driver chip via the I2C or SPI protocol. The LED driver chip performs independent PWM control on each LED, thereby dynamically generating different patterns and text.

5. The two-wheeled electric vehicle intelligent interactive charging system of claim 1, wherein, The facial emotion recognition module uses a wide-angle camera with a near-infrared fill light to ensure that clear facial images can be captured even at night; A lightweight deep learning model is deployed on the NPU of the master module for emotion recognition. The lightweight deep learning is trained on the FER-2013 or AffectNet public dataset to recognize basic emotions such as calm, happy, anxious, angry, and surprised, and output discrete emotion labels and a continuous emotion confidence score (0-1).

6. The two-wheeled electric vehicle intelligent interactive charging system of claim 1, wherein, The multi-language voice interaction module includes a ring-shaped microphone array and a speaker. The ring-shaped microphone array is used to realize sound source positioning and environmental noise reduction. The speaker uses a full-frequency waterproof unit to ensure clear sound quality. The multi-language voice interaction module includes an online mode and an offline mode. The online mode specifically includes integrating the SDK of the voice open platform. When the network connection is good, the user's voice data is uploaded to the cloud after being encrypted, and the cloud dialect recognition engine is used for recognition and semantic understanding. The offline mode specifically includes pre-installing an English voice recognition and synthesis engine in the local storage space. When network interruption is detected or continuous cloud recognition fails, the system automatically switches to offline English mode to ensure basic interaction with foreign users.

7. The two-wheeled electric vehicle intelligent interactive charging system according to any one of claims 1 to 6, characterized in that, The independent entity "one-key help" button of the elderly care module uses a physical anti-misoperation, LED-backlit emergency button with an IP67 protection level. The emergency button is a normally open switch signal inside and is connected to the GPIO port of the master module. 8.The two-wheeled electric vehicle intelligent interactive charging system of claim 1, wherein, The working process of the system is as follows: S1, environmental perception and automatic guidance: The dual-spectrum visual perception module continuously captures environmental images, and the master module calculates the average brightness value of the environmental images. When the average brightness value of the environmental images is lower than the preset threshold, it is determined to be a "dark environment". Once it is determined that the current environment is dark and the state is idle, the master module immediately sends an instruction to the ring-shaped LED matrix projection module to control the ring-shaped LED matrix projection module to project a clear green "empty pile" arrow icon to the ground below. At the same time, the background subtraction or YOLO target detection algorithm is used to analyze the video stream in real time to determine whether the charging pile and the adjacent parking space are occupied by vehicles. If the parking space is occupied, the ring-shaped LED matrix projection module projects a red "occupied" icon or no light; S2, user approach and emotion perception: When the dual-spectrum visual perception module detects a heat source entering the charging area, the face emotion recognition thread is triggered to start, and the face emotion recognition module is activated to capture the user's face image. The user's face image is sent to the emotion recognition model on the local NPU for analysis. The master module quantifies and grades the user's emotions according to the emotion labels and confidence scores output by the lightweight deep learning model. Specifically, Level 0: normal interaction mode; Level 1: mark the user and prepare to start gentle pacification; Level 2: immediately trigger advanced pacification mode and notify the power scheduling system; S3, personalized voice interaction and pacification: the speaker of the multi-language voice interaction module plays a welcome speech, while the ring microphone array of the multi-language voice interaction module starts to listen; if the emotion level is Level 1, pacifying words are integrated into the normal operation guide; if the emotion level is Level 2, the user is actively asked: "You look a bit anxious, do you need me to prioritize your processing?" and a short soothing background music may be played; after the user's reply, cloud dialect recognition is prioritized; if the cloud recognition is successful, the same dialect is replied; if the user speaks English or the cloud connection times out, switch to the local English speech recognition and synthesis engine; S4, identity verification and credit check: the voice prompts the user to perform identity verification: "Please swipe your card or scan the QR code; if the facial emotion recognition module preliminarily judges that the user's age is greater than 65 years old through facial feature analysis, it will specially prompt: "It is detected that you may be an elderly user, it is recommended to use an ID card to swipe the card for verification, and exclusive services can be obtained"; after the user swipes the card or scans the code, the 4G / 5G or Ethernet connection is connected to the background user database to obtain the corresponding user's historical credit score; wherein the user's historical credit score is calculated based on whether the past payment is on time and whether the parking behavior is standardized; S5, power dynamic scheduling: the main control module takes the emotion level and historical credit score of the user of this charging pile as two key weight parameters, encrypts and reports to the cloud charging dispatching management center, the cloud charging dispatching management center makes decisions comprehensively according to the user weight and power grid constraints, the cloud charging dispatching management center sorts all online charging requests according to the weight score, and within the total power limit, higher power is preferentially allocated to users with higher weight scores; wherein the user weight score = A*emotion level + B*credit score; A and B are adjustable coefficients; users with high anxiety level and good credit have high weight; power grid constraints refer to the current total load of the power grid and the requests of other charging piles on the same line; S6, charging execution and continuous monitoring: the user connects the charging gun, confirms and starts charging, and the display screen or LED light spot displays the charging progress; if the user is present and waiting, the emotion change can be intermittently detected, and the interactive content can be dynamically adjusted; the dual-spectrum network camera continuously monitors the temperature of the charging interface; once the "one-key help" button is pressed, the main control module immediately interrupts the current task, initiates a video call request to the remote customer service center through the 4G network, and pushes the real-time video stream and user information of the charging pile to the customer service screen.

Images