Modular multifunctional intelligent trolley and teaching method and system based on intelligent trolley

Abstract

The application discloses a modular multifunctional intelligent trolley and a teaching method and system based on the intelligent trolley, and belongs to the field of education technology and artificial intelligence. The application content comprises a modular multifunctional intelligent trolley, which adopts a double-layer structure composed of a lower load-bearing bottom plate and an upper printed circuit board, integrates a main control, power management and a standardized interface, supports flexible expansion of multi-source sensing and independent vision modules, and the teaching method comprises a four-stage spiral process of perception, cognition, creation and application, guides students to complete full closed-loop learning from image acquisition, labeling, model training to deployment application, constructs a learner ability model through acquisition of behavior data, realizes personalized teaching feedback, and integrates hardware teaching aids, a visual programming environment, course management and a learning community to form a hardware-software collaborative teaching ecological system, and effectively improves the systematicness and practicality of artificial intelligence education.

Description

Technical Field

[0001] This invention belongs to the interdisciplinary field of educational technology and artificial intelligence, specifically relating to a modular multifunctional intelligent vehicle and a teaching method and system based on the intelligent vehicle. Background Technology

[0002] With the rapid development of embedded control, environmental perception, and artificial intelligence technologies, intelligent vehicles have become an important carrier in the field of intelligent mobile robots, widely used in various scenarios such as education and training, intelligent inspection, logistics and distribution, environmental monitoring, and security patrol. Currently, common intelligent vehicles mainly suffer from the following two shortcomings in their overall structural design: One type of product adopts a highly integrated, closed-loop design, with each functional module encapsulated internally. The physical and electrical boundaries between modules are blurred, and the interfaces lack a unified standard, making it extremely inconvenient for functional expansion, module replacement, and system maintenance. This limits its flexible reconfiguration and rapid deployment in different application scenarios.

[0003] Another type of product, while possessing some modular features, suffers from a relatively simplistic design in power management and multi-module collaborative power supply. This makes it difficult to stably and simultaneously provide matching power support for the vision module, multiple sensors, and communication units, impacting the overall operational reliability of the system under complex tasks. Furthermore, existing intelligent vehicles lack systematic consideration in the layout planning of sensors and vision units. The installation positions, angles, and structural relationships do not fully take into account the physical basis of multi-source information fusion, reducing the overall efficiency of the perception system.

[0004] At the educational application level, existing intelligent vehicle teaching platforms mostly remain at the stage of simple programming control or pre-set function demonstration, failing to touch upon the core technical processes of artificial intelligence, especially deep learning. The teaching process lacks a complete closed-loop design from data collection, annotation, model training to deployment and application; course content is fragmented, making it difficult to form a systematic and tiered ability development system. Students often "know what" but not "why," unable to deeply understand the complete lifecycle and collaborative working mechanisms of artificial intelligence systems through hardware platforms.

[0005] Therefore, there is a lack of intelligent vehicle platforms and their supporting teaching methods and systems that can meet the requirements of stability, scalability and reliability in general mobile robot scenarios, and support systematic, closed-loop artificial intelligence teaching processes. Summary of the Invention

[0006] The technical problem to be solved by this invention is: how to provide a modular intelligent vehicle platform and its supporting teaching methods and systems that have a clear structure, well-defined modules, unified interface standards, are easy to expand and maintain, and can support the entire process of artificial intelligence teaching and practice from data acquisition to model deployment, so as to meet the dual needs of general mobile robot applications and systematic teaching and training.

[0007] To solve the above problems, the present invention adopts the following technical solution: In a first aspect, the present invention provides a modular multi-functional intelligent vehicle, including a chassis module, a drive execution module, a power supply module, a main control module, a multi-source sensing module, and a vision module; The chassis module includes a lower load-bearing base plate and an upper functional mounting plate; The drive execution module is installed on the lower load-bearing base plate; The upper functional mounting plate is fixed above the lower load-bearing base plate by a support structure; The power module and the main control module are mounted on the upper functional mounting board; The multi-source sensing module is electrically connected to the upper-level functional mounting plate through a standard interface provided on the upper-level functional mounting plate; The vision module is independently mounted on the upper functional mounting plate via a bracket and is connected to the main control module via a communication interface.

[0008] Optionally, the upper functional mounting plate is a printed circuit board; the printed circuit board integrates a main control module mounting area, a power management circuit, and multiple standardized interfaces for connecting the multi-source sensing module; the standard interfaces on the printed circuit board are pin headers or female headers with a 2.54mm pitch; the printed circuit board has an arc-shaped groove for accommodating the drive wheel; the printed circuit board has a voltage conversion circuit.

[0009] Optionally, the multi-source sensing module includes an ultrasonic ranging module, an infrared tracking module, an infrared obstacle avoidance module, and a voice interaction module; the ultrasonic ranging module is located at the front of the vehicle body, the infrared tracking module is located at the bottom of the vehicle body, and the infrared obstacle avoidance module is located at the front or side of the vehicle body. The drive execution module includes a DC geared motor, a drive wheel, and a motor drive circuit; the motor drive circuit is disposed on the printed circuit board and is electrically connected to the main control module through the traces on the printed circuit board. The power module includes a replaceable battery assembly.

[0010] Optionally, the support structure is a nylon stud; the bracket of the vision module is a long screw stud.

[0011] Secondly, the present invention provides a teaching method based on an intelligent vehicle, applied to the modular multifunctional intelligent vehicle described above, wherein the teaching method based on the intelligent vehicle includes: Perception step: The vision module of the intelligent vehicle is used to collect image data of the target object; Cognitive steps: In a visual deep learning programming environment, a convolutional neural network model is trained based on the image data to obtain a recognition model; Creation steps: Deploy the recognition model to the smart car and write a control program in the programming environment. The control program is used to enable the smart car to perform actions based on the output of the recognition model. Application steps: Run the intelligent vehicle equipped with the aforementioned recognition model and control program in the task scenario and obtain feedback on the running results; The perception step, cognition step, creation step, and application step constitute a cyclical process.

[0012] Optionally, it also includes the step of building a learner competence model: During the cognitive and creative steps, learners' operational behavior data are collected. Based on the operational behavior data, a learner capability model is constructed using deep learning technology; Based on the learner ability model, the difficulty parameters of tasks in the teaching process can be dynamically adjusted or personalized guidance can be provided.

[0013] Optionally, constructing the learner ability model includes: Model learners’ programming serialization behavior data and analyze their logical construction patterns; Clustering algorithms are used to analyze the correlation between learners' hyperparameter selection behavior during model training and the performance of the obtained model, and to classify the learning strategy types. By integrating multi-source behavioral data and project outcome data, a hybrid regression and classification model is constructed using deep neural networks to quantitatively assess and predict learners' ability dimensions.

[0014] Optionally, the cognitive steps specifically include: Learners are guided to annotate image datasets through a visual interface; It offers a variety of pre-built convolutional neural network model architectures for learners to choose from; It provides a graphical interface for learners to adjust model training hyperparameters and displays the changes in training loss and accuracy curves in real time; After the model training is completed, a model performance evaluation report containing metrics such as precision and recall is generated.

[0015] Optionally, the cyclical process is executed repeatedly in different teaching projects, and in each iteration, the complexity of image data acquisition, the complexity of the network structure of the recognition model, and the logical complexity of multi-sensor information fusion in the control program are increased sequentially.

[0016] Thirdly, the present invention provides a teaching system based on an intelligent vehicle for executing the teaching method based on an intelligent vehicle as described above, the system comprising: The intelligent vehicle teaching aid module is used to collect image data and perform physical actions and movements according to the control program. The course management and content delivery module is used to implement the four-stage course process of perception, cognition, creation, and application. The Visual Deep Learning Programming and Simulation Module provides a graphical block-based programming and basic code editing environment, and integrates a simplified deep learning framework, supporting model training, visualization of the training process, and model export. The learning assessment and feedback module is used to collect and analyze learners' operational behavior data based on deep learning technology to build learner ability models and provide personalized learning assessment and feedback. The project showcase and community interaction module is used to display learners' project results and support peer review.

[0017] The beneficial effects of this invention are: 1. Addressing the issues of enclosed structures, ambiguous module boundaries, and inconvenient expansion and maintenance in existing intelligent vehicles, this invention achieves physical separation between the drive execution module and the control and sensing module through a layered design of a lower load-bearing base plate and an upper functional mounting plate. By highly integrating the standardized interfaces of the main control module, power management circuit, and multi-source sensing modules onto the upper functional mounting plate (printed circuit board) and utilizing onboard wiring for internal connections, complex external wiring is significantly reduced, resulting in a hardware platform with a clear structure, reliable connections, and a high degree of modularity. This design not only facilitates the demonstration of circuit structure and system collaboration in teaching scenarios but can also be directly applied to general mobile robot scenarios such as inspection, logistics, and security, demonstrating excellent scenario adaptability and ease of expansion and maintenance.

[0018] 2. To address the issues of simplistic power management and unstable power supply for multiple modules in existing intelligent vehicles, this invention integrates a power management circuit, including a battery input interface and voltage conversion circuit, into the upper functional mounting board. This provides multiple stable voltage outputs, simultaneously offering matched and reliable power support to the main control module, vision module, and various sensors. This effectively ensures the long-term operational stability and compatibility of the system under multi-module and complex task collaborative operation.

[0019] 3. To address the problem of insufficient systematic consideration and limited perception efficiency in the layout of sensors and vision units in existing intelligent vehicles, this invention optimizes the physical layout of multi-source perception by independently mounting the vision module on the vehicle body via a bracket to obtain a wide field of view, and setting ultrasonic, infrared and other sensing modules in specific positions such as the front and bottom of the vehicle body according to their functional zones. This lays a good hardware foundation for the comprehensive collection and fusion processing of environmental information and improves the overall efficiency of the perception system.

[0020] 4. Addressing the issue that existing teaching platforms only involve simple programming control and fail to delve into the core technological processes of artificial intelligence, this invention proposes a teaching method that constructs a four-stage spiral execution process of "perception-cognition-creation-application." This process guides learners to fully participate in the entire process from image data acquisition, annotation, model training, model deployment to intelligent decision-making execution. It also presents core aspects such as deep learning model training and evaluation in a visual and graphical manner, transforming abstract artificial intelligence principles into concrete and operable practical tasks. This effectively solves the shortcomings of fragmented course content and the disconnect between theory and practice, systematically cultivating learners' computational thinking and engineering practice abilities.

[0021] 5. To address the lack of personalized guidance and competency assessment in existing teaching methods, this invention introduces the step of constructing learner competency models into the teaching method. It collects operational behavior data of learners in programming, model training, and other processes, and uses deep learning technology for analysis and modeling. This enables dynamic assessment and prediction of learners' knowledge acquisition, practical ability, and innovative potential, thereby providing data support for dynamic adjustment of teaching content and personalized feedback, and realizing individualized instruction.

[0022] 6. The teaching system provided by this invention organically integrates modular intelligent vehicle hardware, a four-stage spiral curriculum system, a visual deep learning programming environment, a data-driven learning assessment module, and a project demonstration community, forming a complete ecosystem of hardware and software collaboration and a closed loop of teaching, practice, and assessment. This system not only supports the effective implementation of the aforementioned teaching methods but also stimulates learners' creativity and collaborative spirit through community interaction, significantly improving the systematic nature, interactivity, and effectiveness of AI introductory education.

[0023] In summary, this invention provides a highly reliable, easily expandable, and versatile intelligent mobile platform applicable to multiple scenarios through innovative hardware structure design. Simultaneously, by integrating a teaching method and system deeply bound to this hardware platform and incorporating a complete artificial intelligence development process and personalized assessment, it creates a teaching solution that enables a deep understanding and practical application of core artificial intelligence technologies, thus simultaneously meeting the dual needs of general robot applications and high-quality artificial intelligence education and training. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is a three-dimensional view of the intelligent vehicle provided by the present invention.

[0026] Figure 2 This is a top view of the upper PCB board of the intelligent vehicle provided by the present invention.

[0027] Figure 3 The images show the front view and side view of the intelligent vehicle provided by this invention.

[0028] Figure 4 A schematic diagram of the overall architecture of the intelligent teaching system provided by the present invention.

[0029] Figure 5 A flowchart of the intelligent teaching method provided by the present invention.

[0030] Figure 6 This is a schematic diagram of the hardware system composition of the intelligent mobile platform module provided by the present invention.

[0031] Figure 7 The data flow sequence diagram for constructing the learner ability model in the learning assessment and feedback module provided by this invention.

[0032] Figure 8 A schematic diagram of the software interface for the visualization deep learning programming and simulation module provided in this invention.

[0033] Figure 9 This is a schematic diagram illustrating the training process of a convolutional neural network model in the cognitive stage provided by the present invention.

[0034] Figure 10 A schematic diagram of the steps of the four-order spiral teaching process provided by the present invention.

[0035] in Figure 1 and Figure 3 The numbers in the diagram represent the following meanings: 1—Ultrasonic ranging module; 2—Vision module; 3—Infrared obstacle avoidance module; 4—WiFi module; 5—Upper functional mounting plate; 6—Lower load-bearing base plate; 7—Drive wheel; 8—Power module; 9—Main control module. Detailed Implementation

[0036] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] Example 1: Modular Multifunctional Intelligent Vehicle This embodiment provides a modular multi-functional intelligent vehicle, the overall structure of which is as follows: Figure 1 , Figure 3 and Figure 6 As shown, the intelligent vehicle in this embodiment, through clear modular division and structural design, has overall dimensions of 350mm × 250mm × 250mm and a ground clearance of 45mm. The base plate is equipped with multiple standard mounting holes, achieving high reliability, easy expandability, and wide applicability to various scenarios. The intelligent vehicle's structure includes: a chassis module, a drive execution module, a power supply module, a main control module, a multi-source sensing module, and a vision module. This intelligent vehicle is not only designed for teaching scenarios but also considers the need for expansion into other application scenarios during the structural design phase. Specifically, the overall vehicle size, ground clearance, and module installation methods have all been comprehensively designed to adapt to different usage requirements. The multiple standard mounting holes and reserved interfaces on the base plate can be used to expand different types of sensor modules, communication modules, and execution modules, thereby improving the overall adaptability to different task scenarios and secondary development capabilities.

[0038] The chassis module adopts a two-layer structure design (i.e., the intelligent vehicle is constructed in two layers from bottom to top), see [link / reference]. Figure 1 and Figure 3 The lower layer is the lower load-bearing base plate 6, and the upper layer is the upper functional mounting plate 5. The lower load-bearing base plate 6 is preferably made of lightweight and sturdy materials such as acrylic. The drive execution module, including two DC geared motors and drive wheels 7 connected to them, is directly fixed to its lower surface. The upper functional mounting plate 5 is suspended and fixed above the lower load-bearing base plate 6 by a support structure. In this embodiment, the support structure consists of four nylon studs that pass through the mounting holes at the four corners of the upper functional mounting plate 5 and are fixedly connected to the lower load-bearing base plate 6. The lower load-bearing base plate 6 has corresponding threaded holes for nuts, thus forming a stable suspended space between the two layers, achieving both physical isolation and facilitating wiring and heat dissipation. This design makes the overall vehicle structure readily apparent, greatly reducing the complexity of system understanding and maintenance. It is suitable not only for teaching demonstrations but also for rapid maintenance and module replacement in industrial scenarios such as inspection and logistics.

[0039] In this embodiment, the upper functional mounting board 5 is a printed circuit board made of FR-4 material, and its shape is a rectangular flat plate structure. Figure 2As shown, the printed circuit board houses the main electrical modules, excluding the drive execution module. Four symmetrical arc-shaped recesses are formed on both sides of the printed circuit board. The size of these recesses is slightly larger than the drive wheel 7, accommodating the drive wheel 7 of the intelligent vehicle. This design ensures sufficient space for wheel rotation while effectively preventing mechanical interference between the moving parts and the circuit board. This structural feature makes the circuit board itself an integral part of the vehicle structure, optimizing space utilization.

[0040] The power module 8 is mounted on the rear of the upper functional mounting plate 5. It is fixed to the rear of the board via a slot or similar structure and includes a removable and replaceable battery pack. Physically, the battery module uses an 18650 lithium battery pack with a removable battery compartment containing spring-loaded positive and negative contacts. The positive and negative terminals of the battery pack are directly connected to the pre-set power input interface on the printed circuit board via wires or spring-loaded contacts. The innovation of this area lies in the integration of a complete power management circuit. This circuit includes a battery input interface, a voltage conversion circuit, and a multi-channel regulated output circuit. The output voltage of the battery pack is converted by this circuit to stably output 5V and 3.3V. The 5V voltage powers the motor drive circuit and some sensors; the 3.3V voltage powers the main control module 9 and some low-power sensors. These two power supplies are distributed across the rear area of ​​the board via a power layer formed by wide copper traces on the inner layer of the circuit board, providing low-impedance and stable power delivery to each module. This centralized power management design solves the problems of poor power supply matching and susceptibility to interference when multiple modules work together, ensuring the stable operation of high-power units such as vision modules and multiple sensors, and improving the reliability of the system under complex tasks.

[0041] The main control module 9 is installed in the central area of ​​the upper functional mounting board 5. The main control module 9 is fixed to the main control module mounting area on the board via a socket or soldering. Auxiliary circuits necessary for its normal operation are arranged around the main control module 9, including a crystal oscillator, a reset circuit, and interfaces for program debugging. The main control module 9 is electrically connected to the power supply module 8, the drive execution module, and other functional modules via pre-designed etched copper foil traces inside the printed circuit board, thereby achieving centralized control of the entire system.

[0042] The multi-source sensing module connects to the upper-layer functional mounting plate 5 via multiple standardized interfaces. The upper-layer functional mounting plate 5 has several standardized interfaces with 2.54mm pitch pin headers or female headers distributed along its edges and at preset locations. For example... Figure 1 and Figure 2As shown, the ultrasonic ranging module 1 is mounted high on the front of the vehicle body using long screw posts; the infrared obstacle avoidance module 3 is located on the front or side of the vehicle body; the infrared tracking module is mounted on the bottom of the vehicle body facing the ground; the voice interaction module is preferably an independent functional board integrating a microphone array and a voice recognition chip, connected to the mounting board 5 via a 2.54mm pin header interface, used to receive voice commands and / or output voice prompts. The voice interaction module is accessed via an interface. The power, ground, and signal terminals of these sensor modules are fixed with screws and connected to corresponding pins on the upper functional mounting board 5 via 2.54mm pitch pin header interfaces, and then connected to the designated IO port of the main control module 9 via internal wiring, achieving plug-and-play and reliable connection.

[0043] The vision module 2 is independently mounted above the upper functional mounting plate 5 via a bracket. In this embodiment, the vision module 2 is fixed at a height using long screw posts, allowing its camera assembly to obtain a wider forward field of view, facilitating the acquisition of environmental image information. The vision module 2 itself includes a camera and a vision processing unit. It connects to the corresponding socket on the upper functional mounting plate 5 via a dedicated high-speed communication interface, thereby establishing communication with the main control module 9 to achieve image data transmission and feedback of processing results. This module supports running lightweight deep learning models and can receive user-defined models, realizing end-to-end processing from image acquisition to intelligent recognition. Its independent and detachable structure facilitates demonstration, replacement, or upgrades during teaching, while also meeting the flexible adjustment needs of visual perception devices for different application scenarios.

[0044] The motor drive circuit of the drive execution module is also integrated on the upper functional mounting board 5. Specifically, the mounting pads for the motor drive chip and peripheral circuitry are arranged in the front area of ​​the board. This motor drive circuit receives control signals from the main control module 9 through on-board wiring, and connects to the DC geared motor mounted on the lower load-bearing base plate 6 via wires through its output port, thereby driving the wheels 7 to move and realizing the forward, backward, and turning movements of the vehicle. This design replaces the traditional DuPont wire connection between the motor drive board and the main control board with on-board wiring, significantly reducing the number and clutter of external connections, fundamentally avoiding operational instability caused by loose connections and poor contact, and improving the overall reliability of the electrical system.

[0045] To enhance expandability, the upper functional mounting board 5 also features a universal expansion board interface socket and a dedicated interface for the WiFi module 4 in its central area. The dedicated interface for the WiFi module 4 is located in the serial communication interface in the middle of the upper functional mounting board. The interface uses a 2.54mm pitch pin header and includes a 3.3V power supply terminal, a ground terminal, a transmitting terminal (TX), and a receiving terminal (RX). The WiFi module connects to the dedicated interface via a plug-in connection to achieve wireless communication.

[0046] In summary, the intelligent vehicle in this embodiment achieves separation of drive and control through a two-layer physical structure consisting of a lower load-bearing base plate and an upper printed circuit board functional mounting plate. The highly integrated electrical design of the upper printed circuit board integrates power management, main control, drive interfaces, and various standardized sensor interfaces, utilizing onboard wiring instead of external flying wires. Standardized 2.54mm pitch pin headers / female headers enable flexible plug-in of functional modules. The independently elevated vision module and partitioned sensor modules optimize the physical structure of the perception system. These combined features in shape and construction constitute a modular intelligent vehicle platform with a clear structure, reliable connections, convenient expansion, and applicability to various mobile robot application scenarios.

[0047] Example 2: An Interactive Artificial Intelligence Teaching Method Based on Smart Cars This embodiment provides a teaching method based on the modular multifunctional intelligent vehicle described in Embodiment 1. The core of this method lies in constructing a four-stage spiral teaching process of "perception-cognition-creation-application," such as... Figure 5 and Figure 10 As shown, this approach aims to address the fragmentation and disconnect from core technology processes in existing AI teaching. It achieves a comprehensive framework for a deep learning four-stage spiral teaching process by fully participating in the AI ​​system lifecycle. This teaching method revolves around an intelligent vehicle hardware platform, systematically dividing the execution process of a teaching project into four stages: perception, cognition, creation, and application. These four stages form a closed loop and are cyclically used in teaching projects of varying difficulty and themes. Each loop is not a simple repetition but rather presents a spiral increase in data complexity, model complexity, and system integration complexity, thereby achieving a step-by-step development of learners' abilities. This framework transforms abstract deep learning and computer vision technologies into concrete and operable practical tasks that align with the cognitive characteristics of children aged 6-12. It visualizes and makes the principles of learning actionable.

[0048] To clearly demonstrate the application of this method in projects of varying complexity, the following explanations will use basic and advanced projects as examples.

[0049] 1. Example of a basic project: Intelligent color sorting vehicle This project is geared towards beginners and aims to help them understand the basic process and single attribute recognition.

[0050] (1) Perception stage: The teaching objective is to understand data acquisition. The teacher guides students to operate the vision module of the smart car and collect image data of red, yellow and blue blocks under controlled lighting conditions. Students are required to take pictures from different angles and distances so that they can get a preliminary feel for the workload of data acquisition and understand that data is the foundation for subsequent model training.

[0051] (2) Cognitive Stage: The teaching objective is to experience simple model training. Students upload the collected image data to a visualization deep learning platform. The platform automatically divides the dataset into training and validation sets. Students select a three-class classification model from a pre-built library of simplified convolutional neural network models. In the graphical interface, students only need to adjust two hyperparameters: the learning rate and the number of training epochs, to start training. During training, the interface dynamically displays the changes in training loss and validation accuracy using real-time graphs. After training, the platform generates a model performance evaluation report, showing the model's accuracy on the test set and visualizing some successful and unsuccessful recognition cases. The teacher guides students to analyze the reasons for simple misidentifications.

[0052] (3) Creation Stage: The teaching objective is to achieve functional integration. Students will export the best trained model and deploy it to the vision module of the smart car. After completing model training, students can automatically export the best model and convert it into a deployment format supported by the embedded vision module. Depending on the hardware resources, further optimizations such as model compression and quantization can be performed, and finally deployed to the vision module of the smart car to achieve real-time recognition. Subsequently, a simple conditional branch control program is written in the graphical programming environment. The logic is as follows: when the vision module recognizes the result as "red", control the robotic arm or a specific action module to move the red block to area A; when it is recognized as "yellow", move it to area B; when it is recognized as "blue", move it to area C.

[0053] (4) Application Stage: The teaching objective is to complete the test and demonstration. Students test the operation of the cart on a simulated sorting line, observe its recognition and execution accuracy, and record the problems encountered. Finally, the group demonstrates its color sorting system and submits a brief project report.

[0054] 2. Advanced Project Example: Intelligent Traffic Sign Recognition and Response System This project is designed for learners with some prior knowledge and aims to improve their comprehensive application skills in complex scenarios, embodying a spiral progression.

[0055] (1) Perception Stage: The teaching objective is to master complex data acquisition. After learning about traffic signs, students operate intelligent vehicles in groups to collect hundreds of images of common traffic signs, such as stop, left turn, right turn, and speed limit, in simulated road scenarios under different lighting conditions, angles, and even while moving. This allows students to deeply understand the complexity and diversity of real-world data, which is key to building robust models.

[0056] (2) Cognitive Stage: The teaching objective is to gain a deeper understanding of model training and evaluation. This stage consists of three parts: First, students use the platform's bounding box annotation tool to accurately annotate the bounding boxes of the collected traffic sign images to ensure data quality.

[0057] Secondly, such as Figure 9 As shown, students can choose one of several predefined lightweight object detection model architectures and start training after adjusting parameters such as batch size and learning rate. The front end displays the training loss versus validation accuracy curves in real time and introduces validation set performance monitoring, allowing students to intuitively perceive model training, convergence, and potential overfitting.

[0058] Finally, after training is complete, the platform generates a detailed model performance evaluation report. The report not only displays metrics such as precision, recall, and F1 score on the reserved test set, but also visualizes successful and unsuccessful cases using a confusion matrix. Instructors guide students to analyze the reasons for misidentification, such as confusing similar-shaped signs, thereby delving into core concepts such as model generalization ability, overfitting, and underfitting.

[0059] The evaluation indicators are calculated based on the following basic formula: Accuracy: ACC = (TP + TN) / (TP + TN + FP + FN); Precision: Precision = TP / (TP + FP); Recall = TP / (TP + FN); F1 value: F1_score= 2 × Precision × Recall / (Precision + Recall).

[0060] Where TP represents the number of positive classes predicted as positive, TN represents the number of negative classes predicted as negative, FP represents the number of negative classes predicted as positive, and FN represents the number of positive classes predicted as negative.

[0061] (3) Creation Stage: The teaching objective is to master the integration and debugging of complex systems. Students deploy the trained optimal model to the car's vision module. In an environment that supports mixed programming of graphical and Python code, students write complex control programs that include loops, state judgments, and multi-sensor information fusion. For example, the program logic is: when a "left turn sign" is detected, the servo motor is controlled to turn left and the left indicator light is illuminated; at the same time, the program needs to continuously read the ranging data from the ultrasonic sensor. If a nearby obstacle is detected during the turning process, the turning is paused and the obstacle avoidance subroutine is started. Students need to repeatedly debug the program logic and model recognition threshold in a simulated scenario until the system is stable and reliable.

[0062] (4) Application Stage: The teaching objective is to improve comprehensive literacy and expression skills. Each group will demonstrate their intelligent transportation system in a more complex simulated traffic scenario, explain the technical solutions, and answer questions. Each student needs to submit a complete project report, including project concept, dataset description, model selection and training experience, program logic diagram, problems encountered and solutions, summary and outlook. Through writing reports and defense, students will consolidate their knowledge and improve their systematic thinking and expression skills.

[0063] By comparing the beginner and advanced projects, a spiral progression is clearly visible: data acquisition evolves from controlled environments to varied scenarios; model tasks progress from simple image classification to object detection; programming logic evolves from single-conditional branches to complex systems involving loops and multi-sensor fusion; and cognitive levels progress from perceiving specific attributes to understanding abstract rules and system optimization. This fully demonstrates that this method is a scalable and progressively capable teaching framework.

[0064] Example 3: Teaching System Based on Intelligent Cars This embodiment provides a teaching system based on a smart car for implementing the teaching method described in Embodiment 2. For example... Figure 4 As shown, this system is an organic whole that deeply integrates software and hardware, including the following core modules: 1. Intelligent Car Teaching Aid Module The core hardware carrier of this module is the modular multi-functional intelligent vehicle described in detail in Example 1. Its high-performance vision module, multi-source sensor array, stable power management, and clear modular structure provide a highly adaptable and practical hardware platform for teaching. This hardware not only serves as a functional execution unit but also as a teaching demonstration window for students to understand circuit integration, signal flow, and multi-sensor collaboration.

[0065] 2. Course Management and Content Delivery Module This module is responsible for organizing and implementing a four-stage spiral curriculum system of "perception-cognition-creation-application". It includes graded curriculum resource libraries designed for different age groups: 6-8 years old and 9-12 years old. For example, the beginner curriculum library includes projects such as color sorting and object tracking, while the advanced curriculum library includes projects such as traffic sign recognition and dynamic obstacle avoidance. The module has a dynamic progress tracking function, which can record the learning progress of each learner or group at each stage. The content is presented in various formats, including animations, interactive videos, and virtual experiments, to introduce knowledge in a lively and interesting way and stimulate learning interest.

[0066] 3. Visualized Deep Learning Programming and Simulation Module like Figure 8As shown, this module serves as a crucial bridge connecting theoretical understanding with practical creation. It provides a hybrid environment of graphical block-based programming and basic Python code editing, lowering the barrier to entry for programming. Its core innovation lies in the integration of a simplified deep learning framework, specifically optimized for teaching. (1) Data annotation tools: The built-in professional image annotation tools support three annotation methods: rectangle, polygon and key point, and have automatic annotation assistance function.

[0067] (2) Model building and training: A graphical interface is provided, allowing learners to drag and drop components such as convolutional layers, pooling layers, batch normalization layers, and dense layers to build custom convolutional neural network structures. The training interface is extremely simplified, with key hyperparameters such as learning rate and training epochs available for adjustment.

[0068] (3) Visualization of the training process: Real-time plotting and display of curves such as training loss and validation accuracy, dynamic updating of the confusion matrix, making the training process "visible".

[0069] (4) Model management and deployment: After training, a model performance evaluation report containing accuracy, recall, F1 score and confusion matrix visualization can be generated with one click. The model can also be exported in a format that is compatible with the vision processing chip of the smart car, so as to achieve seamless deployment from simulation to real hardware.

[0070] 4. Learning Assessment and Feedback Module like Figure 7 As shown, this module is the core of the system's intelligent implementation. It runs the learner ability model construction and analysis algorithm described in Example 2. This module continuously collects multi-dimensional behavioral data of learners from the visual programming module and the course management module, uses deep learning algorithms for modeling and analysis, and outputs quantitative assessment and prediction reports on learners' knowledge mastery, practical proficiency, and innovative thinking ability. Based on this report, the system can provide early warnings or personalized teaching suggestions to teachers, and can also directly provide adaptive prompts and challenge tasks to learners.

[0071] 5. Project Showcase and Community Interaction Module This module aims to foster a positive learning environment and inspire creativity and collaboration. It provides a child-friendly online community platform where learners can showcase their smart car project results through text, images, videos, or 3D models, such as color sorting demonstration videos or traffic sign recognition system reports. The platform includes peer review, likes, and comments to promote communication, inspiration, and cooperation among learners. Outstanding projects are featured on the homepage, creating positive incentives.

[0072] In summary, the system in this embodiment deeply integrates dedicated hardware teaching aids, a systematic curriculum, visual programming and AI training tools, a data-driven intelligent assessment system, and an incentive-based community platform, constructing a complete teaching ecosystem that spans from hardware operation to software programming, from individual learning to community interaction, and from knowledge transfer to skills development. This effectively supports the spiral-like advancement of early AI education.

[0073] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. 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 of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A modular multi-functional intelligent vehicle, characterized in that, It includes a chassis module, a drive execution module, a power supply module, a main control module, a multi-source sensing module, and a vision module; The chassis module includes a lower load-bearing base plate and an upper functional mounting plate; The drive execution module is installed on the lower load-bearing base plate; The upper functional mounting plate is fixed above the lower load-bearing base plate by a support structure; The power module and the main control module are mounted on the upper functional mounting board; The multi-source sensing module is electrically connected to the upper-level functional mounting plate through a standard interface provided on the upper-level functional mounting plate; The vision module is independently mounted on the upper functional mounting plate via a bracket and is connected to the main control module via a communication interface.

2. The modular multi-functional intelligent vehicle according to claim 1, characterized in that, The upper functional mounting board is a printed circuit board; the printed circuit board integrates a main control module mounting area, a power management circuit, and multiple standardized interfaces for connecting the multi-source sensing module; the standard interfaces on the printed circuit board are pin headers or female headers with a 2.54mm pitch; the printed circuit board has an arc-shaped groove for accommodating the drive wheel; the printed circuit board has a voltage conversion circuit.

3. The modular multi-functional intelligent vehicle according to claim 1, characterized in that, The multi-source sensing module includes an ultrasonic ranging module, an infrared tracking module, an infrared obstacle avoidance module, and a voice interaction module; the ultrasonic ranging module is located at the front of the vehicle body, the infrared tracking module is located at the bottom of the vehicle body, and the infrared obstacle avoidance module is located at the front or side of the vehicle body. The drive execution module includes a DC geared motor, a drive wheel, and a motor drive circuit; the motor drive circuit is disposed on the printed circuit board and is electrically connected to the main control module through the traces on the printed circuit board. The power module includes a replaceable battery assembly.

4. The modular multi-functional intelligent vehicle according to claim 1, characterized in that, The support structure is a nylon stud; the support for the vision module is a long screw stud.

5. A teaching method based on an intelligent vehicle, characterized in that, Applied to the modular multifunctional intelligent vehicle as described in any one of claims 1 to 4, the teaching method based on the intelligent vehicle includes: Perception step: The vision module of the intelligent vehicle is used to collect image data of the target object; Cognitive steps: In a visual deep learning programming environment, a convolutional neural network model is trained based on the image data to obtain a recognition model; Creation steps: Deploy the recognition model to the smart car and write a control program in the programming environment. The control program is used to enable the smart car to perform actions based on the output of the recognition model. Application steps: Run the intelligent vehicle equipped with the aforementioned recognition model and control program in the task scenario and obtain feedback on the running results; The perception step, cognition step, creation step, and application step constitute a cyclical process.

6. The teaching method based on an intelligent vehicle according to claim 5, characterized in that, It also includes the steps of building a learner competence model: During the cognitive and creative steps, learners' operational behavior data are collected. Based on the operational behavior data, a learner capability model is constructed using deep learning technology; Based on the learner ability model, the difficulty parameters of tasks in the teaching process can be dynamically adjusted or personalized guidance can be provided.

7. The teaching method based on an intelligent vehicle according to claim 6, characterized in that, Constructing the learner ability model includes: Model learners’ programming serialization behavior data and analyze their logical construction patterns; Clustering algorithms are used to analyze the correlation between learners' hyperparameter selection behavior during model training and the performance of the obtained model, and to classify the learning strategy types. By integrating multi-source behavioral data and project outcome data, a hybrid regression and classification model is constructed using deep neural networks to quantitatively assess and predict learners' ability dimensions.

8. The teaching method based on an intelligent vehicle according to claim 5, characterized in that, The cognitive steps specifically include: Learners are guided to annotate image datasets through a visual interface; It offers a variety of pre-built convolutional neural network model architectures for learners to choose from; It provides a graphical interface for learners to adjust model training hyperparameters and displays the changes in training loss and accuracy curves in real time; After the model training is completed, a model performance evaluation report containing metrics such as precision and recall is generated.

9. The teaching method based on an intelligent vehicle according to claim 5, characterized in that, The recurring process is executed repeatedly in different teaching projects, and in each iteration, the complexity of image data acquisition, the complexity of the network structure of the recognition model, and the logical complexity of multi-sensor information fusion in the control program are increased sequentially.

10. A teaching system based on an intelligent vehicle, characterized in that, The system is used to perform the teaching method based on a smart car as described in any one of claims 5 to 9, the system comprising: The intelligent vehicle teaching aid module is used to collect image data and perform physical actions and movements according to the control program. The course management and content delivery module is used to implement the four-stage course process of perception, cognition, creation, and application. The Visual Deep Learning Programming and Simulation Module provides a graphical block-based programming and basic code editing environment, and integrates a simplified deep learning framework, supporting model training, visualization of the training process, and model export. The learning assessment and feedback module is used to collect and analyze learners' operational behavior data based on deep learning technology to build learner ability models and provide personalized learning assessment and feedback. The project showcase and community interaction module is used to display learners' project results and support peer review.

Images