Multi-modal perception drag teach handle and interactive control method
By using rigid pillars and six-dimensional force sensing technology, combined with inertial sensing and optimized logic container software, the problems of large size, cumbersome operation and safety hazards of traditional teaching pendants have been solved, achieving an efficient and safe drag-and-drop teaching experience.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- JIANGSU HAUTO NC TECH CO LTD
- Filing Date
- 2026-03-09
- Publication Date
- 2026-06-05
AI Technical Summary
Traditional teach pendants are bulky and cumbersome to operate, cannot be dragged directly, pose safety hazards when switching gaze, lack environmental force perception due to rigid connections, and have complex and time-consuming hardware development.
Rigid columns are used to transmit drag force, combined with six-dimensional force perception and inertial perception, and the logic container software architecture is optimized to achieve blind operation and efficient drag-and-drop teaching.
It achieves an efficient and safe drag-and-drop teaching experience, avoids circuit board damage, reduces hardware dependence, and supports blind operation and instant interface switching.
Smart Images
Figure CN122142958A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of industrial robot auxiliary equipment technology, specifically relating to a multimodal sensing drag-and-drop teaching handle and interactive control method. Background Technology
[0002] In the teaching process of industrial robots, the teach pendant (teach box) is the core device for human-machine interaction. Traditional teach pendants are usually large and connected to the control cabinet via flexible cables. When performing drag-and-drop teaching (i.e., directly and manually pulling the robot's end effector to a designated location), the operator typically needs to hold the teach pendant with one hand to record key presses while using the other hand to push the robot. This operation method is not only inefficient, but also, because the teach pendant is connected via a flexible cable, it cannot be directly used as a force point to drag the robot, resulting in a fragmented operating experience.
[0003] In addition, existing drag handles still have the following significant problems in practical applications:
[0004] Existing controllers typically have a limited number of buttons or rely excessively on touchscreens. When faced with multi-level menus or numerous parameter lists, switching focus one by one using physical buttons is cumbersome; while touchscreens are difficult to operate in oily environments or when wearing gloves. More seriously, during drag-and-drop teaching, the operator's gaze is usually focused on the contact point between the robot's end effector and the workpiece. If the controller needs to be operated (such as recording points), the operator must shift their gaze back to the screen to confirm button triggering. This gaze shift disrupts the continuity of operation and poses a safety hazard.
[0005] Traditional handles often use plastic shells that bear the direct force. During high-intensity dragging operations, the shell is prone to deformation, causing the internal circuit board (PCB) to be squeezed, which can lead to loose solder joints or damage to components.
[0006] Rigidly connected handles lack independent sensing of environmental forces. When the robot end effector experiences an accidental collision, the rigid connection directly transmits the impact force to the robot joints. Typically, the robot cannot stop until the current loop alarm of the robot controller is triggered, resulting in a delayed response and potential damage to the workpiece or equipment.
[0007] Traditional solutions for implementing graphical interfaces often rely on complex Linux kernel-level display driver development, involving cumbersome device tree configuration and driver compilation, which has high hardware requirements and a long development cycle. Summary of the Invention
[0008] Purpose of the invention: The purpose of this invention is to address the shortcomings of existing technologies by providing a multimodal sensing drag-and-drop teaching handle and interactive control method. It achieves reliable transmission of physical drag force through a rigid column, combines six-dimensional force sensing and inertial sensing technologies with an optimized logic container software architecture, and realizes an efficient, safe teaching experience with blind operation capabilities.
[0009] Technical solution: The multimodal sensing drag-and-drop teaching handle of the present invention includes:
[0010] The handle housing assembly includes an upper shell and a lower shell that interlock to form an internal receiving cavity;
[0011] A rigid column is axially mounted within the internal receiving cavity. The rigid column includes a support beam located in the middle and connecting flanges integrally formed at both ends of the support beam. The connecting flanges are respectively engaged in the limiting slots at both ends of the handle housing assembly to achieve axial and radial positioning of the column and the housing.
[0012] The six-dimensional force sensing component is mechanically connected to the connecting flange at the top of the rigid column and is configured to be rigidly coupled to the robotic arm flange to collect force and torque data transmitted through the rigid column.
[0013] An interactive module, integrated on the surface of the upper shell, includes at least one physical input unit;
[0014] The control unit includes a circuit board assembly disposed within the internal receiving cavity; at least one outer wall of the support beam along the axial direction is configured with a flat cross-section, and the circuit board assembly is disposed close to or parallel to the flat cross-section;
[0015] The control unit is communicatively connected to the six-dimensional force sensing component and the physical input unit, and is configured to execute the following control logic in parallel:
[0016] Force control auxiliary logic: Reads the data from the six-dimensional force sensing component in real time and generates motion control commands for the robot by combining them with a preset force control algorithm;
[0017] Interaction Logic: An ordered container is established in the storage space of the control unit. Multiple function commands supported by the handle are instantiated as virtual object elements and filled into the ordered container. A global index is defined to point to the current element in the ordered container. In response to the trigger signal of the physical input unit, the value of the global index is modified to perform logical displacement in the ordered container. According to the confirmation signal of the physical input unit, the function command corresponding to the virtual object element pointed to by the current global index is triggered.
[0018] This invention also provides an interactive control method for the above-mentioned multimodal sensing drag-and-drop teaching handle, comprising the following steps:
[0019] Containerization initialization steps: Encapsulate multiple functional instructions into custom class objects, push them into a logical container in a preset logical order, and assign a unique identifier to each object;
[0020] Force data processing steps: Collect raw data from the six-dimensional force sensor, perform gravity compensation processing, and then convert it into motion vector commands for the robot;
[0021] Index navigation steps: Listen to the signals of the physical input units, update the global index variable pointing to the logical container according to the signal type, or directly map the physical location of the signal to a specific object in the logical container;
[0022] Function triggering steps: When a confirmation signal or a specific key closure signal is detected, the preset function code of the object pointed to by the current index is executed.
[0023] Beneficial effects: Compared with the prior art, the advantages of this invention are as follows: By setting a rigid column as the core force-bearing skeleton, the pushing, pulling, and rotating torques generated by the user are transmitted to the robot joints without loss or delay, realizing the physical dragging function of "handle as handle". At the same time, the circuit board assembly is distributed on the outside of the column and does not directly bear the mechanical load, effectively avoiding the compression of precision components by the deformation of the shell.
[0024] The innovative "gravity scroll wheel" function controls the menu scrolling speed by detecting the tilt angle of the handle, solving the problem of cumbersome browsing of long lists using physical buttons. Combined with the haptic feedback module, the movement of UI focus and function triggering are converted into different vibration codes (such as "segmented" vibrations), enabling true "blind operation".
[0025] It fills the gap in the lack of sensory input in rigid handles. Utilizing a built-in accelerometer to monitor impact force in real time, once the combined acceleration after removing gravity exceeds a safety threshold (e.g., 2.5g), the handle triggers an emergency stop command and a violent vibration alarm before the robot controller, effectively protecting the equipment.
[0026] Employing the SPI framebuffer direct write mechanism, the display can be driven without modifying the operating system kernel code. This not only reduces dependence on the hardware platform but also achieves anti-interference reset through a software timer, maintaining extremely high cost-effectiveness and stability.
[0027] The dynamic mapping mechanism based on JSON configuration files enables millisecond-level interface language switching without restarting the program, meeting the immediate needs of industrial sites. Attached Figure Description
[0028] Figure 1 This is a schematic diagram of the overall structure of the multimodal sensing drag-and-drop teaching handle (with screen) in Embodiment 1 of the present invention; Figure 1 In the diagram, 'a' represents a front view. Figure 1 In the diagram, 'b' represents the reverse side.
[0029] Figure 2 This is a schematic diagram of the overall structure of the multimodal sensing drag-and-drop teaching handle (screenless version) in Embodiment 2 of the present invention; Figure 2 In the diagram, 'a' represents a front view. Figure 2 In the diagram, 'b' represents the reverse side.
[0030] Figure 3 This is a cross-sectional view of the multimodal sensing drag-and-drop teaching handle in Embodiment 2 of the present invention;
[0031] Figure 4 This is a schematic diagram of the initialization logic of the virtual control container in this invention;
[0032] Figure 5 This is a schematic diagram of the cursor movement logic based on global index in this invention.
[0033] Figure 6 This is the main flowchart of the user-space-based screen driver in this invention;
[0034] Figure 7 This is a flowchart of the Chinese character display processing and window coloring process in this invention;
[0035] Figure 8 This is a flowchart of the image display processing and format conversion in this invention. Detailed Implementation
[0036] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings, but the scope of protection of the present invention is not limited to the embodiments described.
[0037] Example 1: As Figure 1 , Figure 2 As shown, the multimodal sensing drag-and-drop teaching handle provided by the present invention mainly consists of a handle housing assembly (including an upper shell 101 and a lower shell 102), a rigid column 105, an interactive module (display screen 103 and silicone buttons 104), a six-dimensional force sensing component 108, and an internal circuit board assembly.
[0038] The handle housing assembly includes an upper shell 101 and a lower shell 102 that interlock, forming an internal receiving cavity. The core improvement of this embodiment lies in the rigid connection of the upright column (…). Figure 1 The middle mark is 105. Figure 2 (Marked as 204). This column is made of aluminum alloy and uses a customized "I-beam" or "dumbbell" structure, mainly including the following features:
[0039] End connecting flanges: Both ends of the column are designed with enlarged circular connecting flanges. Correspondingly, the inner walls of the handle housing are molded with corresponding annular limiting grooves at both ends. During assembly, the flanges at both ends of the column are directly embedded into the grooves in the housing. This design not only achieves precise positioning of the column within the housing, but more importantly, when the user applies push or pull force to the housing, the force is directly transmitted to the column flanges through the groove walls, avoiding force transmission through small screw holes, thus greatly improving the structure's shear resistance.
[0040] like Figure 3 As shown, the D-shaped cross-section support beam consists of the middle section connecting the flanges at both ends. Unlike a circular cross-section, one side of this support beam along the axial direction (i.e., the side facing upwards in the diagram) is designed as a flat cut, giving the beam a "D" shape. Space optimization: Handles typically have limited outer diameters for ergonomic purposes. If the column is a circular tube, it occupies the largest central cylindrical space, forcing the PCB board to be placed along the tangent of the tube, resulting in low space utilization. This design, by "flattening" one side of the column, frees up valuable planar mounting space for the PCB board, allowing it to be placed close to the flat cut of the column. This significantly reduces the overall thickness of the handle, making it easier to grip.
[0041] The space between the handle housing and the column houses the circuit board assembly and power and communication cables (such as USB cables and force sensor signal lines), ensuring the neatness and protection of the wiring harness. The column's walls bear all tensile, compressive, shear, and torque loads. When the user holds the housing and drags the robot, the force is transmitted through the housing to the column, and then from the column to the flange. The internal PCB board is only attached to the gaps around the column and does not participate in the force distribution, thus completely solving the problem of circuit board damage caused by housing deformation in traditional handles.
[0042] The control unit integrates a six-dimensional force sensor (such as the MPU6050) as an inertial sensing module, which acquires the spatial attitude (Euler angles) and three-axis acceleration of the handle in real time via the I2C bus. At the same time, a linear vibration motor is mounted on the back of the PCB as a haptic feedback module, driven by the MCU's PWM interface, to provide operation confirmation and alarm feedback.
[0043] The six-dimensional force sensor is bolted to the connecting flange at the top of the column. When the operator pulls the lower or upper shell by holding the handle, the force passes sequentially through: the inner wall of the shell → the limiting groove at the end of the shell → the connecting flange of the column → the support beam of the column → the top flange → the six-dimensional force sensor. In this path, the precision circuit board assembly is suspended or attached to the flat cut side of the column and does not participate in the transmission of force, thus ensuring that the circuit board will not be damaged by bending under high load dragging conditions.
[0044] This controller is available in two configurations:
[0045] like Figure 1 The screen-equipped version shown includes an upper shell 101, a lower shell 102, an LCD display screen 103, silicone buttons 104, a rigid connecting column 105, a USB port 106, a reset button 107, and a six-dimensional force sensor 108, suitable for scenarios requiring real-time parameter viewing.
[0046] like Figure 2 The full-button version shown includes an upper shell 201, a lower shell 202, a rigid connecting column 204, a USB port 205, and a reset button 206. The screen has been removed, and a full button area 203 has been set up, which is suitable for blind operation or use with an external large screen.
[0047] Example 2: Multimodal Interaction Logic and Blind Operation Implementation
[0048] The control unit runs a core interaction logic based on "containers + indexes + perception," such as... Figure 4 and Figure 5 As shown.
[0049] 1. Containerized Build and Data Structure
[0050] During the program initialization phase, the system does not directly manipulate screen pixels, but instead uses object-oriented design principles to construct the logic layer:
[0051] Custom class definition: such as Figure 6 As shown, the system first defines a "virtual button control class," which not only contains regular UI properties but also encapsulates two core data blocks:
[0052] Unique identifier (object_name): A string field (e.g., "btn_save"), which serves as the control's "ID number" in the system. It is used for subsequent multilingual lookups and logical bindings to ensure that it can be uniquely indexed within the container.
[0053] Class member - stored property element: contains the control's text content, default background color, highlighted background color, bound callback function pointer, etc.
[0054] Container population: The system creates an ordered container (List), such as... Figure 4 As shown in the cylinder on the left. When the interface loads, the system instantiates all interactive buttons on the interface as objects of the above classes, and pushes these objects into a List container in sequence according to the logical order of the interface layout (e.g., from top to bottom). At this time, the discrete buttons on the interface are transformed into a linear, ordered logical chain in memory.
[0055] 2. Global Index Navigation Mechanism
[0056] like Figure 5 As shown, the List container, once filled, displays an ordered stack of buttons from "Button 1" to "Button n". The system defines an integer variable, Global Index, as the global cursor pointer.
[0057] Logical mapping: The value of Global Index directly corresponds to the element index (0 to n-1) in the List container. Figure 7 The arrow in the image visually indicates that Global Index is currently pointing to "Button 2".
[0058] Button-driven navigation: When the user presses the physical "up / down" button, the control unit does not calculate the coordinates, but directly performs an increment or decrement operation on the Global Index.
[0059] For example, pressing the "Down" key increments the Index by 1, and the arrow immediately points from "Button 2" to "Button 3".
[0060] Visually synchronized rendering: The UI rendering thread monitors changes in the Global Index in real time. Once the index changes, the system reads the "class member properties" of the new object (such as "Button 3"), renders its background color property as a highlight color (such as green), and simultaneously restores the old object ("Button 2") to its default color (such as white). This mechanism ensures that the user can clearly perceive the current focus of the operation without touching the screen.
[0061] 3. Gravity roller navigation
[0062] To improve long lists (i.e.) Figure 7 To improve browsing efficiency when n is relatively large, the system introduces a gravity navigation mode. When the user presses and holds the "enable button" on the side and tilts the controller, the controller's posture is mapped to... Figure 7 The rate of change of the Global Index is measured when the user presses the "Enable" button on the side and tilts the handle along the vertical axis; the system then reads the pitch angle from the inertial sensing module.
[0063] like Start a low-speed timer (e.g., 2Hz) to automatically trigger index updates and achieve slow scrolling; if A high-speed timer (e.g., 10Hz) is activated to enable rapid page turning. This allows users to quickly browse the parameter list with just a slight turn of their wrist, greatly improving operational efficiency.
[0064] 4. Tactile feedback and blind operation:
[0065] To compensate for the lack of visual focus, the system has established a "visual-touch fusion" mechanism. Whenever the Global Index changes (whether triggered by a button or gravity), the control unit drives the motor to generate a short vibration of 20ms to simulate the "tactile feedback" of a mechanical roller.
[0066] In a screenless, all-button configuration ( Figure 2 In this system, a matrix mapping logic is used: the matrix coordinates (Row, Col) of the physical buttons are directly mapped to the key values (Key) of the container. Operation feedback is achieved through vibration encoding: for example, a successful save of a point will vibrate twice (fast rhythm), while a failure will result in a single long vibration (slow rhythm). The operator does not need to take their eyes off the workpiece; they can confirm the operation result simply by the vibration of their palm.
[0067] Example 3: User-space based screen driver technology
[0068] For gamepads with screens, to reduce hardware costs and avoid complex kernel development, this invention adopts the following... Figure 6 The user space software synchronization scheme shown.
[0069] 1. Framebuffer direct write:
[0070] The system starts a timed loop task (e.g., 30ms period) to directly open the Linux system's frame buffer device / dev / fb0 in user space.
[0071] Text rendering ( Figure 7 ): Obtain the trajectory by looking up the character library based on the string coordinates, and then color the pixel buffer.
[0072] Image rendering ( Figure 8 ): Parses the image header and converts the pixel data into the RGB565 format supported by the screen.
[0073] SPI transmission: The synthesized dirty rectangle data is directly written to the LCD driver chip via the SPI bus.
[0074] 2. Anti-interference reset:
[0075] In the aforementioned cycle, the system periodically checks the screen status register or forcibly sends an initialization command sequence. If static electricity or electromagnetic interference in the industrial environment causes the screen to go blank or registers to become corrupted, this mechanism can automatically repair the display within tens of milliseconds during the next frame refresh, ensuring extremely high reliability.
[0076] Example 4: Active Safety Monitoring
[0077] This embodiment utilizes a built-in inertial sensing module to implement active safety functions.
[0078] Monitoring logic: Outside the main loop, a safe thread runs independently to calculate the magnitude of the synthesized acceleration vector in real time.
[0079] Triggering mechanism: A safety threshold (e.g., 2.5g) is set. Due to the rigid connection, hard collisions at the robot's end effector are transmitted to the handle without attenuation. When the magnitude of the acceleration vector (after removing the influence of gravity) exceeds the safety threshold, an abnormal collision is determined to have occurred.
[0080] Response action: The control unit immediately sends the highest priority "Stop" command to the robot controller through the communication interface; at the same time, the UI thread is suspended, and the drive motor outputs full-speed continuous vibration (alarm mode) until the reset button is detected to be pressed.
[0081] Example 5: Online Hot Switching for Multiple Languages
[0082] To address the issue of needing to restart when switching languages, this invention employs a JSON-based configuration scheme.
[0083] A predefined structured array is used, containing fields such as key (unique identifier), ch (Chinese), and en (English), where the key strictly corresponds to the object_name of the virtual control.
[0084] When a "language switch" command is received, the control unit does not restart the program. Instead, it iterates through each control object within the container, uses its `object_name` to look up the corresponding new language string in a JSON dictionary, and calls the object's `setText()` method to directly overwrite the memory property. The next time the screen refreshes, the interface changes to the new language, achieving a seamless switch within milliseconds.
[0085] As described above, although the invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the invention itself. Various changes in form and detail may be made without departing from the spirit and scope of the invention as defined in the appended claims.
Claims
1. A multimodal sensing drag-and-drop teaching handle, characterized in that, include: The handle housing assembly includes an upper shell and a lower shell that interlock to form an internal receiving cavity; A rigid column is axially mounted within the internal receiving cavity. The rigid column includes a support beam located in the middle and connecting flanges integrally formed at both ends of the support beam. The connecting flanges are respectively engaged in the limiting slots at both ends of the handle housing assembly to achieve axial and radial positioning of the column and the housing. The six-dimensional force sensing component is mechanically connected to the connecting flange at the top of the rigid column and is configured to be rigidly coupled to the robotic arm flange to collect force and torque data transmitted through the rigid column. An interactive module, integrated on the surface of the upper shell, includes at least one physical input unit; The control unit includes a circuit board assembly disposed within the internal receiving cavity; at least one outer wall of the support beam along the axial direction is configured with a flat cross-section, and the circuit board assembly is disposed close to or parallel to the flat cross-section; The control unit is communicatively connected to the six-dimensional force sensing component and the physical input unit, and is configured to execute the following control logic in parallel: Force control auxiliary logic: Reads the data from the six-dimensional force sensing component in real time and generates motion control commands for the robot by combining them with a preset force control algorithm; Interaction Logic: An ordered container is established in the storage space of the control unit. Multiple function commands supported by the handle are instantiated as virtual object elements and filled into the ordered container. A global index is defined to point to the current element in the ordered container. In response to the trigger signal of the physical input unit, the value of the global index is modified to perform logical displacement in the ordered container. According to the confirmation signal of the physical input unit, the function command corresponding to the virtual object element pointed to by the current global index is triggered.
2. The multimodal sensing drag-and-drop teaching handle according to claim 1, characterized in that, The cross-section of the supporting beam is D-shaped, and the flat section is the planar area on the outer wall of the supporting beam. The flat cut surface is positioned opposite to the inner surface of the upper shell to provide clearance space for electronic components on the circuit board assembly; The internal cavity is provided with a cable channel, and the connecting flange is provided with a cable outlet hole communicating with the cable channel for accommodating the cable connecting the six-dimensional force sensing component and the control unit.
3. The multimodal sensing drag-and-drop teaching handle according to claim 1, characterized in that, The control unit is also configured to execute a gravity compensation algorithm: The spatial attitude of the handle is obtained using an inertial measurement unit integrated on the circuit board assembly; Based on the overall mass distribution model of the rigid column and handle, the gravity component of the handle itself is calculated in combination with the spatial posture. The gravity component is removed when processing the data from the six-dimensional force sensing component to achieve zero-force drag teaching of the robot.
4. The multimodal sensing drag-and-drop teaching handle according to claim 1, characterized in that, The handle is configured as a fully button-operated blind terminal, the physical input unit is a multi-row, multi-column button matrix, and the handle does not include an embedded display screen; The ordered container is configured as a two-dimensional mapping container; The control unit is configured to execute spatial mapping logic: directly mapping the physical row and column coordinates of the key matrix to the index keys of the two-dimensional mapping container; When a user presses a designated button in the button matrix, the control unit directly locates the corresponding virtual object element according to the mapping relationship and drives the handle to generate a vibration signal of a specified mode to confirm the triggering of the function command.
5. The multimodal sensing drag-and-drop teaching handle according to claim 1, characterized in that, The handle is configured as a visual interactive terminal and also includes a display screen embedded in the surface of the upper shell; The display screen is located in the area above the physical input unit; The control unit is also configured to execute view-touch synchronized rendering logic: Based on the value of the global index, the currently selected virtual object element is identified, and the display screen is driven to highlight the visual representation of the virtual object element. The control unit operates in user space and periodically reads pixel data from the system display cache, converts it into RGB format, and writes it to the display screen through a serial peripheral interface to achieve real-time interface refresh.
6. The multimodal sensing drag-and-drop teaching handle according to claim 1, characterized in that, The control unit is also equipped with a dynamic language mapping mechanism: A pre-stored structured configuration file containing multiple sets of "key name - multilingual text" mappings; When a language switching command is received, all virtual object elements in the ordered container are traversed, the object name of each object is extracted as the key, a matching item is searched in the structured configuration file, and the matched target language text is assigned to the virtual object element in real time. The language switching on the display screen can be completed without restarting the system.
7. The multimodal sensing drag-and-drop teaching handle according to claim 1, characterized in that, The control unit is also configured to execute safety monitoring logic: Set acceleration safety thresholds and force control safety thresholds; When the acceleration of the handle's movement or the detected external force exceeds the safety threshold, the control unit immediately interrupts the motion control commands sent to the robot and triggers a high-frequency alarm vibration.
8. An interactive control method applied to the multimodal sensing drag-and-drop teaching handle according to any one of claims 1-7, characterized in that, Includes the following steps: Containerization initialization steps: Encapsulate multiple functional instructions into custom class objects, push them into a logical container in a preset logical order, and assign a unique identifier to each object; Force data processing steps: Collect raw data from the six-dimensional force sensor, perform gravity compensation processing, and then convert it into motion vector commands for the robot; Index navigation steps: Listen to the signals of the physical input units, update the global index variable pointing to the logical container according to the signal type, or directly map the physical location of the signal to a specific object in the logical container; Function triggering steps: When a confirmation signal or a specific key closure signal is detected, the preset function code of the object pointed to by the current index is executed.
9. The method according to claim 8, characterized in that, The method also includes a screen driving step: Based on the updated global index variable, determine the virtual object element that is currently selected and needs to be highlighted in the ordered container; The system frame buffer device is directly opened and read in the user space of the operating system. Based on the coordinate position and size attributes of the virtual object element, the image data of the dirty rectangular region is synthesized in memory. The image data is sent to the display driver chip in real time via a serial peripheral interface, and the haptic feedback module is triggered to vibrate while the data is being sent, so as to achieve synchronization between visual feedback and haptic feedback.
10. The method according to claim 8, characterized in that, The method also includes a collision emergency stop procedure: The composite vector magnitude of the acceleration data is calculated in real time. When the synthesized vector magnitude exceeds a preset safety threshold, an abnormal collision is determined, and the current logical mapping step is immediately interrupted to stop updating the global index. Send the highest priority motion stop command to the robot controller via an external communication interface; Simultaneously, the tactile feedback module is driven to perform a high-frequency continuous vibration alarm until a physical reset signal or a system restart signal is detected.