A small four-axle unmanned aerial vehicle sensing and control interaction system for teaching

Abstract

The application discloses a small four-axle unmanned plane sensing and control interaction system for teaching, which comprises a fusion sensing module, an integrated flight control and electronic governor integrated board and a fuselage cover which is integrally detachable. The fusion sensing module is internally arranged on the fuselage cover and comprises a binocular camera, an ultrasonic ranging sensor and a local data preprocessing subunit, is used for directly collecting environmental image information and near-distance information in a teaching scene, and preliminarily completes noise reduction filtering and data labeling preprocessing locally. The flight control and electronic governor integrated board is connected with the fusion sensing module through a direct hardware interface to realize point-to-point local communication. The application realizes architecture optimization through integrated design of the flight control and electronic governor, local structured preprocessing of sensing data, self-defined short-distance direct communication protocol and multi-sensor dynamic weight fusion decision, greatly reduces the hardware failure rate of the educational unmanned plane in the teaching scene, gets rid of the dependence on the cloud network, and comprehensively improves the scene adaptability and classroom use experience of the educational unmanned plane teaching aid.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) technology, and more specifically to a sensory control and interactive system for a small quadcopter UAV used for teaching. Background Technology

[0002] With the rapid popularization of artificial intelligence and aviation science education in primary and secondary schools, educational drones have become one of the core teaching tools in science and technology innovation classrooms for teenagers. However, current educational drone solutions in the industry still have many shortcomings in adaptability and cannot fully meet the actual needs of normalized teaching scenarios. The main deficiencies of educational drones currently on the market are as follows: First, the hardware architecture is redundant, resulting in a high failure rate in teaching systems. Traditional teaching drones generally adopt a separate layout of "independent sensing module + separate flight control board + external ESC module". A large number of external cables are required between different modules to complete signal transmission. In the teaching scenario where teenagers repeatedly disassemble and move the drones, problems such as loose cables, poor contact, and signal interference occur frequently. On average, the monthly failure rate of every 10 teaching tools exceeds 30%, which significantly reduces the limited hands-on teaching time in the classroom and also increases the maintenance cost of teaching tools for schools.

[0003] Secondly, there is redundancy in perceived data, resulting in poor adaptability for teaching in offline environments. Most solutions on the market rely on transmitting all perception data, such as raw images and point clouds, back to the cloud for recognition processing. This not only consumes extremely high transmission bandwidth but also requires a stable wireless network environment to operate normally. However, the local area network environment of most primary and secondary school classrooms and laboratories is complex, and some remote school areas even lack stable external network coverage. This directly causes the scene recognition function of drones to fail, making it impossible to complete preset teaching demonstrations such as obstacle avoidance and marker following.

[0004] Meanwhile, the single-sensor solution has obvious drawbacks and is not well-suited for teaching scenarios: Currently, mainstream teaching drones are either equipped with only ultrasonic sensors, which cannot distinguish between obstacle types and are prone to misjudging objects such as transparent glass and thin gauze; or they are equipped with only monocular / dual-eye vision sensors, which are prone to inaccurate ranging in special environments such as classroom lighting changes frequently, backlighting, and complex ground textures. The obstacle avoidance misjudgment rate of both types of solutions generally exceeds 40% in typical teaching scenarios, which can easily cause collisions and damage to teaching aids and also pose certain classroom safety hazards.

[0005] Finally, the rigid strategies fail to meet the needs of a tiered teaching approach throughout the entire learning cycle. Most existing obstacle avoidance strategies for teaching drones adopt fixed threshold designs, which cannot be dynamically adjusted according to the learner's proficiency. In the beginner stage, the obstacle avoidance intervention is insufficient, which easily leads to accidental collisions. However, after the learner has mastered the basic operation, the excessive system intervention will limit them from completing high-difficulty training actions such as flying through circles and flying close to the ground. This makes it difficult to adapt to the teaching pace of the whole process from zero-based beginner to advanced skill improvement and to support the design of tiered science and technology innovation courses.

[0006] To address the existing pain points in the industry, this solution proposes a multi-sensor fusion data interaction system for educational scenarios. Through integrated architecture design, local edge preprocessing, and dynamic weight allocation, it fully adapts to the specific usage needs of primary and secondary school teaching scenarios. Summary of the Invention

[0007] The purpose of this invention is to provide a sensory control and interactive system for a small quadcopter drone used in teaching, aiming to solve the problems and defects in the aforementioned background technology. Specifically, this invention adopts the following technical solution: The present invention provides a small quadcopter drone sensor and control interactive system for teaching purposes, comprising: Fusion perception module: Integrated inside the detachable body cover, it includes a binocular camera, an ultrasonic ranging sensor, and a local data preprocessing subunit. It is used to directly collect environmental image information and close-range information in teaching scenarios, and to complete preliminary noise reduction filtering and data labeling preprocessing locally. Flight control and electronic speed control integrated board: It achieves point-to-point local communication with the fusion sensing module through a direct hardware interface. This integrated board integrates the flight control unit and electronic speed control unit on the same PCB board, and can directly process sensing data and generate power adjustment commands without cross-board forwarding. Ground control terminal: Wirelessly connected to the flight control ESC integrated board, with a built-in obstacle avoidance parameter adjustment module, supporting teachers or students to visually select teaching modes corresponding to different proficiency levels, and dynamically distributing perception data interaction rules and obstacle avoidance trigger thresholds.

[0008] Furthermore, the local data preprocessing subunit of the fusion perception module also has a built-in teaching-specific recognition model, which can complete the identification of teaching aids and the classification and labeling of obstacles in the classroom before data is uploaded. It can compress the original image data into labeled structured feature data before transmitting it to the flight control and electronic speed controller integrated board.

[0009] Furthermore, the flight control ESC integrated board and the fusion perception module adopt a custom short-range direct connection communication protocol, which integrates the traditional multiple sets of parallel cables of multiple peripherals into a single set of dual-channel transmission lines, and transmits pre-processed ranging data, structured visual feature data and sensor calibration control commands sent back from the flight control terminal.

[0010] Furthermore, the flight control and electronic speed control integrated board has a built-in fusion decision subunit, which is used to simultaneously receive two types of data from the fusion perception module, cross-validate the accurate distance value returned by ultrasonic ranging with the obstacle contour features returned by binocular vision, and eliminate invalid data points generated by a single sensor due to environmental interference.

[0011] Furthermore, the flight control and electronic speed control integrated board has a built-in dynamic threshold mapping table, which corresponds one-to-one with the three teaching modes of beginner, intermediate and advanced levels issued by the ground control terminal. Different obstacle avoidance data trigger weights are automatically matched in different modes. The beginner mode prioritizes the use of high-weight ultrasonic data to ensure safety, while the advanced mode prioritizes the use of visual feature data to open up more operating space.

[0012] In addition, this application also discloses a data interaction method for a sensor-control interaction system of a small quadcopter drone used for teaching, including the following steps: S1 Hardware Direct Connection Initialization: After the UAV is powered on, the fusion perception module sends a hardware identity verification signal to the flight control ESC integrated board through the direct connection interface to complete sensor ID registration and communication channel bandwidth pre-allocation; S2 local preprocessing of perception data: The fusion perception module simultaneously collects binocular visual images and ultrasonic ranging data, performs noise filtering and invalid frame removal locally, and calls the built-in teaching-specific recognition model to complete obstacle classification and labeling, transforming the raw data into lightweight structured feature data; S3 Point-to-Point Low-Latency Transmission: The fusion perception module transmits the labeled structured feature data to the flight control and electronic speed controller integrated board in real time through a single direct connection channel. The frame sequence check code is attached during the transmission process to ensure data integrity. S4 Data Fusion Decision: After receiving two types of perception data, the flight control and electronic speed control integrated board retrieves the dynamic threshold mapping table corresponding to the current teaching mode for cross-verification. When the judgment results of the two types of data are consistent and the obstacle avoidance trigger threshold of the current mode is reached, the power adjustment control command is directly generated in the board. S5 Power Command Direct Drive Output: The integrated flight control and electronic speed control board does not require cross-board forwarding. It directly outputs control commands to the four coreless motors through the onboard electronic speed control unit to perform corresponding deceleration, hovering, obstacle avoidance and other actions. S6 Mode Dynamic Iteration: After receiving the mode switching command from the teacher or student, the ground control terminal wirelessly sends the updated dynamic threshold configuration parameters to the flight control and electronic speed control integrated board, updating the fusion decision rules in real time to adapt to the operational needs of different learning stages.

[0013] Furthermore, in the data fusion decision-making process of step S4, if there is a deviation between the judgment results of ultrasonic ranging data and binocular visual feature data, the system will prioritize marking the area as a potential obstacle, execute a deceleration warning action, and synchronously upload the deviation data to the ground control terminal to assist teachers in assessing the teaching scenario environment.

[0014] Furthermore, the integrated sensing module supports hot-swappable disassembly. During disassembly, the flight control and electronic speed controller integrated board can automatically identify offline signals from the sensing module and switch to the basic flight mode that relies solely on the inertial measurement unit, ensuring minimum operational reliability in basic operation teaching scenarios.

[0015] Furthermore, the data interaction latency is controlled within 20ms, and the perception data does not need to be forwarded across boards from acquisition to power command output. It is adapted to the high-speed response characteristics of the hollow cup motor of micro UAV, meeting the high-speed response requirements of sudden misoperation and emergency stop and dynamic obstacle avoidance in teaching scenarios.

[0016] Furthermore, the teaching-specific recognition model can directly identify the unique logos of the drone's accompanying teaching aids, completing teaching task feedback without uploading the original images to the cloud, reserving computing power for advanced teaching functions such as visual programming and path recognition.

[0017] Because the present invention adopts the above technical solution, it has the following beneficial effects: 1. High reliability and low redundancy in hardware architecture: By integrating the flight controller and electronic control into the same integrated board, and using point-to-point direct transmission for the fusion sensing module, the multiple sets of external cables caused by traditional multiple peripherals are completely eliminated. The internal space occupied by the whole machine is reduced by more than 60%, and the failure rate of cable detachment is reduced to less than 10% of that of traditional split architecture, which is fully adapted to the requirements of low maintenance cost in the teaching scenario for teenagers.

[0018] 2. Improved perception data fusion: Through a two-level processing mechanism of local preprocessing and on-board fusion verification, the complementary advantages of ultrasonic ranging being unaffected by lighting and binocular vision being able to identify obstacle types are fully utilized. The obstacle avoidance misjudgment rate is reduced by 85% in special scenarios such as sudden changes in classroom lighting and multi-machine cluster training, effectively avoiding problems such as collisions with walls or machines due to misoperation during training.

[0019] 3. Significantly enhanced teaching adaptability: The built-in dynamic threshold mapping table is linked with the mode adjustment module of the ground terminal, which can switch the obstacle avoidance strategy to adapt to the three learning stages of beginner, intermediate and proficient with one click, realizing a smooth transition from "high safety and full protection" to "low intervention and high openness", meeting the step-by-step teaching needs of teenagers.

[0020] 4. Significantly reduced control response latency: Data interaction does not require cross-board forwarding throughout the entire process. The end-to-end latency of sensing data from acquisition to power command output is controlled within 20ms, which greatly improves the response speed to sudden misoperation and emergency stop in teaching scenarios and dynamic obstacle avoidance, and fully adapts to the high-speed response characteristics of the hollow cup motor of micro UAV.

[0021] 5. Enhanced Functionality: The integrated perception module's locally pre-recognized teaching-specific model can directly identify the unique identifiers of the accompanying teaching aids, completing teaching task feedback without uploading the original images to the cloud, reserving computing power for subsequent advanced teaching functions such as visual programming and path recognition. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the drone hardware of the present invention; Figure 2 This is a hardware connection block diagram of the present invention; Figure 3 This is a flowchart of the multi-sensor fusion decision-making logic of the present invention; Figure 4 This is a schematic diagram illustrating the data weight allocation in the teaching model of this invention. Figure 5 This is a flowchart illustrating the data interaction timing of the present invention. Detailed Implementation

[0023] To enhance understanding of the present invention, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0024] like Figures 1 to 5 As shown, the interactive sensory control system for teaching small quadcopter drones of the present invention includes: Fusion perception module: Integrated inside the detachable body cover, it includes a binocular camera, an ultrasonic ranging sensor, and a local data preprocessing subunit. It is used to directly collect environmental image information and close-range information in teaching scenarios, and to complete preliminary noise reduction filtering and data labeling preprocessing locally. Flight control and electronic speed control integrated board: It achieves point-to-point local communication with the fusion sensing module through a direct hardware interface. This integrated board integrates the flight control unit and electronic speed control unit on the same PCB board, and can directly process sensing data and generate power adjustment commands without cross-board forwarding. Ground control terminal: Wirelessly connected to the flight control ESC integrated board, with a built-in obstacle avoidance parameter adjustment module, supporting teachers or students to visually select teaching modes corresponding to different proficiency levels, and dynamically distributing perception data interaction rules and obstacle avoidance trigger thresholds.

[0025] Based on the above features, in the overall design of the primary and secondary school teaching drone, this application fixes the fusion perception module inside the detachable cover on the top of the fuselage using a snap-fit ​​structure. The overall size of the module is controlled within 30mm×25mm×12mm, and it is integrally formed with the cover without the need for additional screws. The flight controller and ESC integrated board adopts a 35mm diameter circular PCB design, which arranges the STM32F411 main controller, four-channel hollow cup ESC drive circuit, and six-axis IMU sensor on the same board layer. It directly plugs into the fusion perception module through the onboard 2.54mm pitch direct plug interface, completely eliminating the traditional external ribbon cable between the flight controller and ESC. The ground control terminal uses a dedicated teaching APP adapted for Android tablets, which establishes a wireless connection with the drone via Bluetooth 5.0. The APP interface has a built-in visual mode selection slider. The system allows teachers to switch control strategies for different teaching scenarios with a single click. The entire hardware solution is directly compatible with micro teaching drones with a wheelbase of 85mm. By designing the integrated sensing module, flight control and ESC integrated board, and ground control terminal in an architecture-level collaborative manner, it completely breaks the traditional hardware layout of teaching drones where "sensing-flight control-ESC" are completely separated. On the one hand, it significantly reduces the number of external cables inside the fuselage, reducing the failure rate of loose cables and signal interference, which are common in teaching scenarios, by more than 90%. On the other hand, it achieves full-link local closed-loop processing of sensing data from acquisition to power output, enabling basic obstacle avoidance decisions to be made without relying on external computing power. It is fully compatible with normal teaching use in primary and secondary school classrooms, laboratories, and other environments without network access, while reserving sufficient hardware expansion space for subsequent functional iterations.

[0026] More specifically, the local data preprocessing subunit of the fusion perception module also has a built-in teaching-specific recognition model, which can complete the identification of teaching aids and the classification and labeling of obstacles in the classroom before data is uploaded. It can compress the original image data into labeled structured feature data before transmitting it to the flight control and ESC integrated board. Based on the above description, the local preprocessing subunit of the fusion perception module is equipped with a low-power K210 edge computing chip. The pre-trained lightweight recognition model specifically for teaching is burned into the chip's built-in Flash memory. The model retains the recognition capabilities of only four core teaching scene targets: desks, human bodies, teaching sign circles, and landmark stickers. The inference time for a single frame image is controlled within 1ms. When the module is running, it automatically filters out irrelevant background information such as trees outside the window and ceiling lights, and compresses the original 640×480 image data into a 32-byte structured data packet containing only the target category ID, distance value, and contour bounding box coordinates, without needing to transmit it to the flight control system. Any raw pixel data significantly reduces bus bandwidth usage. The teaching-specific recognition model is built into the fusion perception module, which can directly complete the recognition of teaching aids and the classification and labeling of obstacles in the classroom before data is uploaded. The raw image data is compressed into lightweight structured feature data. Compared with directly transmitting the full raw image, the data transmission bandwidth usage is reduced by 70%. This not only avoids a large amount of redundant data crowding out the computing resources of the flight control system, but also filters out invalid interference information that is irrelevant to the teaching scenario in advance. This allows the flight control system to directly obtain perception results with scene attributes, which greatly improves the recognition accuracy of typical teaching obstacles such as desks, human bodies, and teaching signs.

[0027] More specifically, the flight control ESC integrated board and the fusion perception module adopt a custom short-range direct communication protocol, which integrates the traditional multiple sets of parallel cables of multiple peripherals into a single set of dual-channel transmission lines, and transmits pre-processed ranging data, structured visual feature data and sensor calibration control commands sent back from the flight control terminal.

[0028] Based on the above description, the custom short-range direct-connection communication protocol adopts a half-duplex dual-channel design. The physical layer only requires a 4-pin ribbon cable to complete power supply and bidirectional data transmission, with 2 pins for 5V power supply and the remaining 2 pins for differential signal transmission. The protocol frame format is set to 1 byte frame header, 1 byte data type identifier, 6 bytes payload, and 1 byte parity bit, with a total frame length of only 9 bytes. The transmission baud rate is configured to 2Mbps, and it can stably transmit 30 frames of sensing data per second. It also supports the flight controller to send control commands such as sensor zero-point calibration and exposure parameter adjustment, completely replacing the traditional 6-pin cable. This multi-peripheral connection scheme features ~8 independent cabling units and employs a custom short-range direct-connection communication protocol. It integrates multiple sets of parallel cabling for traditional multi-peripheral devices into a single set of dual-channel transmission lines, reducing the internal wiring space of the fuselage by 60% at the physical level. This allows for further reduction in the overall size of the small teaching drone. Simultaneously, it enables bidirectional synchronous transmission of sensor uplink data and flight control downlink calibration commands, avoiding the command delay issues caused by traditional time-division transmission. The dynamic calibration response speed of the sensors is improved by 50%, allowing for rapid sensor calibration at any time in teaching scenarios without the need for factory returns for debugging, significantly reducing the maintenance cost of teaching aids.

[0029] More specifically, the flight control and electronic speed control integrated board has a built-in fusion decision subunit, which is used to simultaneously receive two types of data from the fusion perception module, cross-validate the accurate distance value returned by ultrasonic ranging with the obstacle contour features returned by binocular vision, and eliminate invalid data points generated by single sensors due to environmental interference.

[0030] Based on the above explanation, during operation of the fusion decision subunit of the flight control and electronic speed control integrated board, ultrasonic ranging values ​​and visual recognition distance results are synchronously collected every 10ms, with a ±20% deviation tolerance threshold set: when the distance difference between the two types of data is within the threshold range, the accurate ultrasonic ranging value is directly used as the final judgment result; when the difference exceeds the threshold, the visual obstacle classification result is given priority. If the visual recognition identifies a target that is easily penetrated by ultrasonic waves, such as transparent glass or thin gauze, an obstacle avoidance warning is directly triggered, completely avoiding misjudgment problems caused by the failure of a single sensor; through The integrated flight control and electronic speed control board has a built-in fusion decision subunit that cross-validates the accurate distance data from ultrasonic ranging with the obstacle contour features from binocular vision. This perfectly compensates for the shortcomings of a single sensor in certain scenarios. It solves the problem of inaccurate ranging in environments with excessively low or high light levels, and avoids the shortcomings of pure ultrasonic ranging, such as the inability to distinguish obstacle types and the tendency to misjudge transparent objects. In special scenarios such as classrooms with frequent lighting changes and complex ground textures, the obstacle avoidance misjudgment rate is reduced from 45% in traditional solutions to less than 5%, making it fully adaptable to the complex environment of youth teaching scenarios.

[0031] More specifically, the flight control ESC integrated board has a built-in dynamic threshold mapping table, which corresponds one-to-one with the three teaching modes of beginner, intermediate and advanced levels issued by the ground control terminal. Different obstacle avoidance data trigger weights are automatically matched in different modes. The beginner mode prioritizes the use of high-weight ultrasonic data to ensure safety, while the advanced mode prioritizes the use of visual feature data to open up more operating space.

[0032] Based on the above explanation, the flight control board's built-in dynamic threshold mapping table pre-sets three sets of configurable parameters: In beginner mode, ultrasonic data accounts for 70% of the weight, visual data accounts for 30%, and the obstacle avoidance trigger distance is set to 1m. Deceleration and hovering are triggered as long as either sensor detects an obstacle within 1m. In advanced mode, both types of data account for 50% of the weight, and the obstacle avoidance trigger distance is set to 0.6m. Obstacle avoidance is only triggered when both types of data simultaneously determine that an obstacle has entered the 0.6m range. In proficient mode, visual data accounts for 70% of the weight, ultrasonic data accounts for 30%, and the obstacle avoidance trigger distance is set to 0.3m. Obstacle avoidance is only triggered for visually recognized hard obstacles, and low-level obstacle avoidance is fully enabled. The system allows for the operation of advanced maneuvers such as low-altitude flight and circling maneuvers. It's important to note that the flight control system incorporates a dynamic threshold mapping table corresponding to each of the three teaching modes. This enables differentiated obstacle avoidance strategies for learners with varying levels of proficiency: In beginner mode, ultrasonic data is given high weight, prioritizing absolute safety and minimizing collision risks for novices; in advanced mode, both types of data are weighted evenly, balancing safety and operational flexibility; and in proficient mode, visual data is given high weight, significantly reducing system intervention and allowing learners to perform advanced maneuvers such as circling and edge-flying. This truly achieves a full-cycle, step-by-step teaching coverage from beginner to advanced skill development.

[0033] In addition, this application also discloses a data interaction method for a sensor-control interaction system of a small quadcopter drone used for teaching, including the following steps: S1 Hardware Direct Connection Initialization: After the UAV is powered on, the fusion perception module sends a hardware identity verification signal to the flight control ESC integrated board through the direct connection interface to complete sensor ID registration and communication channel bandwidth pre-allocation; S2 local preprocessing of perception data: The fusion perception module simultaneously collects binocular visual images and ultrasonic ranging data, performs noise filtering and invalid frame removal locally, and calls the built-in teaching-specific recognition model to complete obstacle classification and labeling, transforming the raw data into lightweight structured feature data; S3 Point-to-Point Low-Latency Transmission: The fusion perception module transmits the labeled structured feature data to the flight control and electronic speed controller integrated board in real time through a single direct connection channel. The frame sequence check code is attached during the transmission process to ensure data integrity. S4 Data Fusion Decision: After receiving two types of perception data, the flight control and electronic speed control integrated board retrieves the dynamic threshold mapping table corresponding to the current teaching mode for cross-verification. When the judgment results of the two types of data are consistent and the obstacle avoidance trigger threshold of the current mode is reached, the power adjustment control command is directly generated in the board. S5 Power Command Direct Drive Output: The integrated flight control and electronic speed control board does not require cross-board forwarding. It directly outputs control commands to the four coreless motors through the onboard electronic speed control unit to perform corresponding deceleration, hovering, obstacle avoidance and other actions. S6 Mode Dynamic Iteration: After receiving the mode switching command from the teacher or student, the ground control terminal wirelessly sends the updated dynamic threshold configuration parameters to the flight control and electronic speed control integrated board, updating the fusion decision rules in real time to adapt to the operational needs of different learning stages.

[0034] Based on the above embodiments, the entire data interaction process is automatically executed after the UAV is powered on: First, after power-on, the fusion perception module completes sensor self-testing within 200ms and sends an ID registration signal to the flight control board, which automatically allocates a dedicated communication bandwidth of 1Mbps. Second, the perception module simultaneously acquires binocular images and ultrasonic data, performs three moving average filters locally, and calls the teaching recognition model to complete target labeling. Third, the structured data is transmitted in real time through a direct connection channel, with each frame accompanied by a CRC8 checksum; transmission failures are automatically retransmitted no more than twice. Fourth, the flight control system retrieves the weight parameters of the current mode to complete fusion verification; after determining that obstacle avoidance is triggered, it directly generates PWM speed control signals for four motors on the board. Fifth, the onboard ESC directly outputs drive signals, completing power control within 20ms. The deceleration action; after the ground APP switches modes in the sixth step, the new weight parameters are sent to the flight controller via Bluetooth, and the parameter hot update is completed within 1 second, taking effect without restarting the drone; this full-process data interaction method, through the complete link design of hardware direct connection initialization, local preprocessing, point-to-point low-latency transmission, in-board fusion decision, direct drive power output, and dynamic mode iteration, controls the end-to-end latency of perception data from acquisition to power command output to within 20ms. Compared with the traditional cross-board transmission solution, the response speed is improved by more than 3 times. In the face of emergencies such as sudden misoperation by students or dynamic obstacles intruding in teaching scenarios, it can achieve millisecond-level emergency stop response. At the same time, it supports seamless switching of teaching modes and can complete strategy updates without restarting the drone, fully ensuring the smoothness of classroom teaching.

[0035] More specifically, during the data fusion decision-making process in step S4, if there is a deviation between the ultrasonic ranging data and the binocular visual feature data, the system prioritizes marking the area as a potential obstacle, executes a deceleration warning, and synchronously uploads the deviation data to the ground control terminal to assist teachers in assessing the teaching scenario environment. When the deviation between the ultrasonic and visual data exceeds a preset threshold, the flight control system does not directly trigger a forced stop, but instead reduces the drone's current flight speed by 30%. At the same time, a yellow warning prompt pops up in the teaching background of the ground APP, marking the potential abnormal area of ​​the current drone position. After class, teachers can export all deviation data to statistically analyze the sensor recognition blind spots in the classroom. The system can adjust the placement of obstacles in the training area based on their distribution location to avoid misjudgment risks during subsequent teaching. It also automatically records such deviation samples and regularly updates the local recognition model via OTA to optimize its adaptability. When discrepancies arise between the two types of perception data, potential obstacles are prioritized for marking, and a deceleration warning is issued. This avoids collisions caused by occasional failures of a single sensor without directly triggering a forced stop that would disrupt the trainee's normal operating rhythm. Simultaneously, the deviation data is uploaded to the ground control terminal, allowing teachers to analyze blind spots in the current teaching environment and adjust the training area layout accordingly, further optimizing the safety of the teaching environment.

[0036] More specifically, the fusion perception module supports hot-swappable overall disassembly. During disassembly, the flight controller integrated circuit board can automatically identify offline signals from the perception module and switch to the basic flight mode relying solely on the inertial measurement unit, ensuring minimum operational reliability in basic operation teaching scenarios. The fusion perception module uses a foolproof snap-fit ​​installation, allowing for manual disassembly without tools. The connection lines between the module and the flight controller board are equipped with hardware detection pins. When the module is pulled out, the flight controller detects a change in pin level within 10ms and automatically switches to the basic flight mode of pure IMU + barometer, disabling all obstacle avoidance logic and retaining only manual remote control. Flight Functionality: In this mode, the drone can be adapted to basic control training courses without sensory input. Even if the sensor module is damaged, it will not directly lose control and crash, greatly improving the fault tolerance and service life of the teaching aid. The integrated sensor module supports hot-swappable overall disassembly. The flight control system can automatically recognize the offline status and switch to the basic flight mode. On the one hand, it is convenient for teachers to quickly replace sensor modules with different functions to adapt to different teaching course content. On the other hand, when the sensor module fails, the drone can still maintain basic flight capability and will not directly lose control and crash, greatly improving the fault tolerance and service life of the teaching aid and reducing the probability of sudden failures in the teaching process.

[0037] The data interaction latency is controlled within 20ms, and the perception data does not need to be forwarded across boards from acquisition to power command output. It is adapted to the high-speed response characteristics of the hollow cup motor of micro UAV, meeting the high-speed response requirements of sudden operation and emergency stop and dynamic obstacle avoidance in teaching scenarios. In sudden teaching scenarios such as student operation errors or dynamic obstacles suddenly entering the classroom, the system can complete emergency stop and obstacle avoidance actions in a very short time, greatly reducing the risk of collision and ensuring the safety of personnel and equipment during the teaching process.

[0038] The teaching-specific recognition model can directly identify the unique logos of drone-equipped teaching aids, completing teaching task feedback without uploading original images to the cloud. It reserves computing power for advanced teaching functions such as visual programming and path recognition. Even in scenarios where classroom WiFi signals are unstable or outdoor science activities are without network access, it can stably complete teaching functions such as teaching aid recognition and task determination, completely avoiding teaching interruptions caused by network dependence.

[0039] In practical applications, young learners can control the drone of this invention for flight training using a handle or programming terminal. The ultrasonic ranging sensor and binocular camera perceive the surrounding environment in real time. When an obstacle is detected at a distance less than a preset safety threshold, the integrated flight control and electronic speed controller (ESC) controls the coreless motor to adjust its speed, performing actions such as hovering, obstacle avoidance, or returning to home. Teachers can access the obstacle avoidance settings interface via the accompanying mobile app or ground station software. In this interface, core parameters such as 'safe distance' and 'obstacle avoidance sensitivity' can be freely adjusted according to the needs of different training stages, enabling tiered teaching. For example, beginners can set a larger safe distance and higher sensitivity, while advanced learners can gradually reduce the safe distance and sensitivity to challenge themselves with more difficult maneuvers. Adjustment commands are sent to the integrated flight control and ESC via a wireless link (such as Bluetooth or 2.4G remote controller data transmission), where its internal processor executes the corresponding obstacle avoidance strategy adjustments. When the battery is low, learners can quickly replace the battery to ensure the continuity of teaching.

[0040] As a preferred embodiment, the binocular camera built into the fusion perception module of this system has a capture resolution of 640*480 and an output frame rate of 30fps. The effective detection distance of the ultrasonic ranging sensor is set to 0.1-3m. The local data preprocessing subunit is equipped with a low-power MCU, which can complete obstacle classification and labeling of a single frame image within 1ms. The original image data is compressed into 32-byte structured data containing only obstacle distance, type, and contour features, which greatly reduces the data transmission bandwidth usage.

[0041] The flight control and electronic speed control integrated board uses the STM32F4 series main controller, integrating the flight control computing unit and four electronic speed control units on a circular PCB board with a diameter of only 35mm. It is directly connected to the fusion sensing module through the onboard dual-channel SPI interface, with a theoretical transmission bandwidth of up to 10Mbps, which fully meets the real-time transmission requirements of lightweight structured data.

[0042] The obstacle avoidance parameter adjustment module of the ground control terminal has three default teaching modes: In the beginner mode, the obstacle avoidance safety threshold is set to 1m, and the decision weight of ultrasonic data accounts for 70%, ensuring absolute safety for beginners; in the advanced mode, the obstacle avoidance safety threshold is set to 0.6m, and the weights of ultrasonic and visual data each account for 50%, balancing safety and operational flexibility; in the proficient mode, the obstacle avoidance safety threshold is set to 0.3m, and the decision weight of visual data accounts for 70%, minimizing system intervention and meeting the training needs of students to complete high-difficulty maneuvers such as flying through circles and skimming along edges.

[0043] In actual teaching scenarios, when the ultrasonic sensor of the fusion perception module detects an obstacle 0.8m ahead, and the binocular camera identifies the obstacle as a classroom desk, the fusion perception module transmits the labeled structured data to the integrated flight control and electronic speed controller (ESC) board. If the system is in introductory mode, the fusion decision subunit directly triggers obstacle avoidance, and the onboard ESC unit immediately outputs a power adjustment signal, causing the drone to automatically decelerate and hover, completely avoiding the risk of collision. Teachers can adjust the mode parameters at any time on the ground terminal. After the configuration change signal is transmitted to the integrated flight control and ESC board via Bluetooth, the new strategy is loaded and takes effect within 1 second, without requiring a drone restart, ensuring the smoothness of classroom teaching.

[0044] In summary, this invention significantly reduces the hardware failure rate of educational drones in teaching scenarios through architectural optimizations such as integrated flight control and electronic speed control design, local structured preprocessing of perception data, customized short-range direct communication protocol, and dynamic weighted fusion decision-making of multiple sensors. It also eliminates dependence on cloud networks, significantly reduces obstacle avoidance misjudgment rate in complex classroom environments, and supports the adaptation to the full-cycle, step-by-step teaching needs from beginner to advanced skill improvement, comprehensively enhancing the scenario adaptability and classroom user experience of educational drone teaching aids.

[0045] The present invention has been disclosed above with reference to preferred embodiments, but it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed structure and technical content to create equivalent embodiments without departing from the scope of the present invention, and all such modifications or alterations shall still fall within the scope of the present invention.

Claims

1. A sensor-control interactive system for a small quadcopter drone used for teaching, characterized in that, include: Fusion perception module: Integrated inside the detachable body cover, it includes a binocular camera, an ultrasonic ranging sensor, and a local data preprocessing subunit. It is used to directly collect environmental image information and close-range information in teaching scenarios, and to complete preliminary noise reduction filtering and data labeling preprocessing locally. Flight control and electronic speed control integrated board: It achieves point-to-point local communication with the fusion sensing module through a direct hardware interface. This integrated board integrates the flight control unit and electronic speed control unit on the same PCB board, and can directly process sensing data and generate power adjustment commands without cross-board forwarding. Ground control terminal: Wirelessly connected to the flight control ESC integrated board, with a built-in obstacle avoidance parameter adjustment module, supporting teachers or students to visually select teaching modes corresponding to different proficiency levels, and dynamically distributing perception data interaction rules and obstacle avoidance trigger thresholds.

2. The interactive sensory control system for a small quadcopter drone used for teaching according to claim 1, characterized in that: The local data preprocessing subunit of the fusion perception module also has a built-in teaching-specific recognition model, which can complete the identification of teaching aids and the classification and labeling of obstacles in the classroom before data is uploaded. The original image data is compressed into labeled structured feature data before being transmitted to the flight control and electronic speed control integrated board.

3. The interactive sensory control system for a small quadcopter drone used for teaching according to claim 2, characterized in that: The flight control and electronic speed controller integrated board and the fusion perception module adopt a custom short-range direct connection communication protocol, which integrates the traditional multiple sets of parallel cables of multiple peripherals into a single set of dual-channel transmission lines, and transmits pre-processed ranging data, structured visual feature data and sensor calibration control commands sent back from the flight control terminal.

4. The interactive sensory control system for a small quadcopter drone used for teaching according to claim 1, characterized in that: The flight control and electronic speed adjustment integrated board has a built-in fusion decision subunit, which is used to simultaneously receive two types of data from the fusion perception module, cross-verify the accurate distance value returned by ultrasonic ranging with the obstacle contour features returned by binocular vision, and eliminate invalid data points generated by a single sensor due to environmental interference.

5. The interactive sensory control system for a small quadcopter drone used for teaching according to claim 4, characterized in that: The flight control and electronic speed control integrated board has a built-in dynamic threshold mapping table, which corresponds one-to-one with the three teaching modes of beginner, intermediate and advanced levels issued by the ground control terminal. Different obstacle avoidance data trigger weights are automatically matched in different modes. The beginner mode prioritizes the use of high-weight ultrasonic data to ensure safety, while the advanced mode prioritizes the use of visual feature data to open up more operating space.

6. A data interaction method based on the system of claim 1, characterized in that, Includes the following steps: S1. Hardware Direct Connection Initialization: After the UAV is powered on, the fusion perception module sends a hardware identity verification signal to the flight control ESC integrated board through the direct connection interface to complete sensor ID registration and communication channel bandwidth pre-allocation. S2. Local preprocessing of perception data: The fusion perception module simultaneously collects binocular visual images and ultrasonic ranging data, performs noise filtering and invalid frame removal locally, and calls the built-in teaching-specific recognition model to complete obstacle classification and labeling, transforming the raw data into lightweight structured feature data. S3. Point-to-point low-latency transmission: The fusion perception module transmits the labeled structured feature data to the flight control and electronic speed controller integrated board in real time through a single direct connection channel. The frame sequence check code is attached during the transmission process to ensure data integrity. S4. Data Fusion Decision: After receiving two types of perception data, the flight control and electronic speed adjustment integrated board retrieves the dynamic threshold mapping table corresponding to the current teaching mode for cross-verification. When the judgment results of the two types of data are consistent and the obstacle avoidance trigger threshold of the current mode is reached, the power adjustment control command is directly generated in the board. S5. Direct Power Command Output: The integrated flight control ESC board does not require cross-board forwarding. It directly outputs control commands to the four coreless motors through the onboard electronic speed control unit to perform corresponding deceleration, hovering, obstacle avoidance and other actions. S6. Dynamic Mode Iteration: After receiving the mode switching command from the teacher or student, the ground control terminal wirelessly sends the updated dynamic threshold configuration parameters to the flight control and electronic speed control integrated board, updating the fusion decision rules in real time to adapt to the operational needs of different learning stages.

7. The data interaction method of a small quadcopter drone sensor and control interaction system for teaching as described in claim 6, characterized in that: In the data fusion decision-making process of step S4, if there is a deviation between the judgment results of ultrasonic ranging data and binocular visual feature data, the system will prioritize marking the area as a potential obstacle, execute a deceleration warning action, and synchronously upload the deviation data to the ground control terminal to assist teachers in assessing the teaching scenario environment.

8. The data interaction method of a small quadcopter drone sensor and control interaction system for teaching as described in claim 6, characterized in that: The fusion sensing module supports hot-swappable overall disassembly. During the disassembly process, the flight control and electronic speed control integrated board can automatically identify the offline signal of the sensing module and switch to the basic flight mode that only relies on the inertial measurement unit, ensuring the minimum operational reliability in basic operation teaching scenarios.

9. The data interaction method of a small quadcopter drone sensor control interaction system for teaching according to claim 6, characterized in that, The data interaction latency is controlled within 20ms, and the perception data does not need to be forwarded across boards from acquisition to power command output. It is adapted to the high-speed response characteristics of the hollow cup motor of micro UAV, and meets the high-speed response requirements of sudden misoperation and emergency stop and dynamic obstacle avoidance in teaching scenarios.

10. The interactive sensory control system for a small quadcopter drone used for teaching according to claim 2, characterized in that, The teaching-specific recognition model can directly identify the unique identifiers of the drone's accompanying teaching aids, completing teaching task feedback without uploading the original images to the cloud, and reserving computing power space for advanced teaching functions such as visual programming and path recognition.

Images