Tunnel blasting charging robot detonator safety clamping and loading force control system and method

Abstract

The application discloses a tunnel blasting charging robot detonator safety clamping and loading force control system and method, and relates to the field of tunnel blasting charging robots. It comprises a tunnel blasting charging robot detonator safety clamping and loading force control system and a force control method based on the force control system. The force control system comprises a robot body, a detonator clamping and loading execution unit and a control unit arranged on the robot body. The force control method comprises visual identification and path planning, multi-stage force control clamping, compliant carrying process, compliant loading process and data recording and self-learning. The application has the beneficial effects that: through multi-stage force control clamping, force / position hybrid compliant loading, full-process collaborative control, multi-sensor fusion and multiple safety redundancies, adaptive, high-precision, high-safety and strong anti-interference operation of detonator clamping and loading are realized, and the operation safety, detonation reliability and construction intelligent level are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent equipment technology for tunnel blasting construction, specifically to a system and method for safe clamping and loading force control of detonators in a tunnel blasting charging robot. Background Technology

[0002] As tunnel construction using the drill-and-blast method moves towards intelligence and automation, the problems of high safety risks, low efficiency, and poor loading accuracy inherent in traditional manual loading operations are becoming increasingly prominent. Tunnel blasting loading robots have emerged as a core piece of equipment to replace manual labor and improve construction safety and efficiency. Among these, the detonator, as a key initiating element in tunnel blasting, directly determines the success or failure of the blasting operation through the safety and stability of its clamping and loading process. Excessive clamping force can easily lead to damage to the detonator's outer shell and compression of the internal explosives, causing an explosion. Insufficient clamping force can result in unstable clamping and detonator slippage. Improper loading force can cause the detonator and explosives to not adhere tightly, affecting the detonation effect, or damage the detonator due to excessive impact. Therefore, the safe clamping of the detonator and adaptive loading force control have become one of the core technical challenges in the development of loading robots.

[0003] Currently, the detonator clamping and loading force control technologies for tunnel blasting loading robots are mainly divided into two categories: one is a control method that uses fixed clamping force and fixed loading force. By preset fixed mechanical clamping force parameters, the robotic arm clamps the detonator and then pushes it into the explosive hole with a constant pushing force; the other is a simple pressure sensor feedback control method. A pressure sensor is installed on the clamp to detect the clamping force in real time. When the clamping force reaches a preset threshold, the clamping stops, and the loading process uses a fixed pushing force or a simple proportional adjustment method.

[0004] Although existing technologies can achieve basic clamping and loading functions of detonators, in actual tunnel construction scenarios, due to factors such as differences in detonator specifications, deviations in the position of explosive holes, dust interference in the construction environment, and errors in the operation of robotic arms, existing technologies have many defects and shortcomings, as detailed below: The lack of adaptability in clamping force control poses significant safety hazards: Existing technologies mostly use fixed clamping force parameters, failing to consider the differences in shell strength between different detonator specifications, and also failing to respond in real time to changes in force during the clamping process. When the preset clamping force is too large, it can easily squeeze the detonator shell, leading to detonator damage, explosive leakage, or even an explosion. When the preset clamping force is too small, stable clamping of the detonator cannot be achieved, and the detonator is prone to slipping and falling into the explosives pile or construction area during the movement and turning of the robotic arm, also posing a significant safety risk. At the same time, some technologies that use simple pressure feedback can only achieve coarse control by "stopping when a threshold is reached," and cannot make real-time fine adjustments based on the dynamic force during the clamping process, still resulting in problems such as unstable clamping or over-clamping. Insufficient precision in controlling the loading force affects blasting effectiveness and poses safety hazards: Existing technologies often use a fixed pushing force in the loading process, without adaptive adjustment based on actual working conditions such as the fit gap between the detonator and the explosive hole, and the looseness of the explosive. If the loading force is too large, it will cause excessive impact force between the detonator and the explosive, which may damage the detonator fuse or squeeze the explosive, affecting the reliability of detonation. If the loading force is too small, the detonator cannot be fully inserted into the bottom of the explosive hole, or it may not fit tightly with the explosive, which may lead to problems such as delayed detonation and incomplete blasting, thereby affecting tunnel excavation efficiency and even causing safety hazards such as residual explosive. The clamping and loading process lacks coordinated control and has low operational precision: In the existing technology, the clamping force control and loading force control are mostly independent control modules, and the coordinated linkage between the two is not realized. During the clamping process, the positioning accuracy of the robotic arm, the matching degree of the clamping angle and the loading force are insufficient. It is easy for the detonator to deviate after clamping, and then rub against the inner wall of the explosive hole during loading, which may damage the detonator or cause the loading position to deviate, further affecting the blasting effect. Weak anti-interference ability and poor adaptability: The tunnel construction environment is complex, with a large amount of dust, vibration, temperature fluctuations and other interference factors. The pressure detection elements in the existing technology are easily affected by dust and vibration, which leads to a decrease in detection accuracy and thus affects the accuracy of clamping force and filling force control. At the same time, the existing technology does not adapt to different specifications of detonators, different types of explosives and different tunnel construction conditions. It has poor versatility and requires frequent manual adjustment of parameters, which reduces work efficiency and increases the safety risks caused by manual intervention. Lack of safety redundancy design and poor fault tolerance: In the existing technology, when the pressure detection element fails, the robotic arm deviates, or the detonator specification is misjudged, the abnormality cannot be identified in time and the emergency control cannot be triggered, which can easily lead to safety accidents. At the same time, there is no real-time monitoring and alarm mechanism for clamping force and loading force. Operators cannot keep track of the force state during the clamping and loading process of the detonator in time, and it is difficult to predict and avoid risks in advance.

[0005] In summary, the existing detonator clamping and loading force control technology of tunnel blasting loading robots has defects such as poor adaptability of clamping and loading forces, insufficient control precision, poor coordination, weak anti-interference ability, and insufficient safety redundancy, which cannot meet the actual needs of safe, accurate, and efficient clamping and loading of detonators in tunnel blasting construction. Summary of the Invention

[0006] To address the aforementioned deficiencies in existing technologies, this invention provides a detonator safety clamping and loading force control system and method for tunnel blasting explosive loading robots. The aim is to solve the technical problems of poor adaptability, insufficient control precision, weak coordination, poor anti-interference capability, and insufficient safety redundancy in the detonator clamping and loading process in existing technologies. This achieves adaptive, high-precision, high-safety, and strong anti-interference operation of the detonator clamping and loading, effectively improving operational safety, detonation reliability, and the level of intelligent construction.

[0007] To achieve the above objectives, the present invention provides the following technical solution: On one hand, the present invention provides a detonator safety clamping and loading force control system for a tunnel blasting explosive loading robot, including a robot body, and a detonator clamping and loading execution unit and a control unit disposed on the robot body.

[0008] Furthermore, the robot body includes a six-axis robotic arm, and a detonator clamping and loading execution unit is detachably mounted at the end of the robotic arm via an actuator quick-change interface; the detonator clamping and loading execution unit includes a clamping mechanism, a multi-dimensional force / torque sensor, a displacement sensor, a vision recognition component, and a proximity sensor; The clamping mechanism includes an electric gripper, and a buffer layer is provided on the inner side of the gripper. The multidimensional force / torque sensor is installed between the gripping mechanism and the end of the robotic arm. It adopts strain gauge or piezoelectric sensor to sense the axial force, radial force and torque of the gripper during gripping, handling and loading in real time. The displacement sensor is installed inside the clamping mechanism or uses an encoder built into the robotic arm joint to monitor the opening and closing degree of the gripper and the precise position of the detonator in space in real time, with a resolution of no more than 0.1 mm. The visual recognition component includes at least one high-definition industrial camera and a supplementary light source, used to identify the posture, model, end face position of the detonator, and the filling port position of the explosive cartridge. The proximity sensor is installed at the front end of the gripper and uses laser ranging or capacitive proximity switch to sense the contact state between the detonator and the cartridge or borehole wall at the end of the filling stage, serving as an auxiliary criterion for force control switching.

[0009] Furthermore, the control unit includes a host computer, a slave computer, and a safety protection module; the host computer is responsible for path planning, force control algorithm calculation, visual image processing, human-computer interaction, and data recording; the slave computer is responsible for receiving instructions and controlling the servo drive of the robotic arm and gripping mechanism in real time with a 1ms cycle to form a high-bandwidth force closed loop; the safety protection module is a hardware-level safety circuit independent of the main control system, which can immediately cut off the drive power and trigger an alarm when the gripping force or loading force exceeds the preset hardware threshold.

[0010] On the other hand, the present invention provides a force control method for the safe clamping and loading of detonators in a tunnel blasting charging robot, which is based on the above-mentioned force control system for the safe clamping and loading of detonators in a tunnel blasting charging robot, and includes the following steps: S1. Visual Recognition and Path Planning: The explosive loading robot collects images of the detonator tray area through the visual recognition unit, identifies the detonator target and calculates its six-degree-of-freedom pose, plans a collision-free clamping path, and moves the gripper to the clamping starting point. S2. Multi-level force control gripping: The robot's robotic arm and end gripper adopt a fuzzy PID adaptive gradient gripping force control algorithm based on online stiffness identification. It executes a progressive control strategy that includes coarse position positioning, online identification of detonator shell stiffness, force / position hybrid fuzzy PID adaptive adjustment, variable gradient incremental loading, early warning of plastic deformation precursor sliding mode observation, and gripping reliability verification to complete the safe gripping of the detonator. S3. Compliant handling: After the robotic arm grips the detonator, it uses an impedance control strategy to plan the handling path, so that the robotic arm has compliance with external collisions during the movement. S4. Compliant loading: When loading detonators, the gripper at the end of the robotic arm adopts a progressive process of visual guidance into the hole, online identification of working conditions and switching of force control mode, predictive compliant advancement of the variable impedance model, intelligent identification and adaptive processing of multi-source abnormal working conditions, multi-feature fusion judgment of the position and gradient compliant release, to load the detonator to the target position. S5. Data Recording and Self-Learning: Record job parameters to build a database and optimize control parameters through offline analysis.

[0011] Furthermore, in step S1, the visual recognition unit acquires images of the detonator tray area using a high-definition industrial camera, identifies the detonator target using a deep learning-based YOLO or MaskR-CNN network, and uses the PnP (Perspective-n-Point) algorithm to calculate the detonator's central axis, end face position, and six-degree-of-freedom spatial pose. Based on the recognition results and combined with the robotic arm's kinematic model, the control system plans a collision-free clamping path.

[0012] Furthermore, in step S2, Coarse positioning: The robotic arm moves the gripper to the outside of the detonator and closes the gripper quickly according to the trapezoidal speed plan in the position control mode. When the flexible buffer layer of the gripper makes slight contact with the detonator shell and the displacement sensor shows a continuous change in the gripper opening value, the system records the current initial opening and initial contact force to complete the preliminary positioning. Online identification of detonator shell stiffness: Based on the force-displacement sampling data during the initial contact stage, the system uses the recursive least squares method to identify the equivalent stiffness and yield critical force of the current detonator shell online. The specific identification logic is as follows: using the regression vector composed of the force sensor sampling value and the displacement sensor sampling value of the current sampling period as input, based on the parameter vector identified in the previous sampling period, and combined with the recursive gain matrix, the parameter identification error of the current period is calculated, and the equivalent stiffness and yield critical force of the detonator shell in the current sampling period are updated in real time. Based on the identified equivalent stiffness and yield critical force, the system adaptively generates the target clamping force and the graded loading strategy. The calculation logic of the target clamping force is: the target clamping force is equal to the product of the detonator mass, gravitational acceleration, the maximum expected acceleration of the robotic arm and the safety factor, and then divided by the static friction coefficient between the gripper and the detonator shell. Force / Position Hybrid Fuzzy PID Adaptive Adjustment: The system switches to a force / position hybrid control mode, using the gripper opening degree as the position control loop and the gripping force as the force control loop to construct a dual-loop coupled fuzzy PID controller; the gripping force deviation and deviation change rate are used as inputs to the fuzzy controller, and the proportional coefficient, integral coefficient, and derivative coefficient of the PID controller are self-tuned online through preset fuzzy rules. The fuzzy rules adopt the Mamdani inference model of "if the deviation and deviation change rate are equal, then the proportional coefficient, integral coefficient, and derivative coefficient are equal". The centroid method is used for defuzzification to achieve overshoot-free and fast-response control of the gripping force; Variable gradient incremental loading: Based on the online identification of the detonator shell yield critical force, the target clamping force is divided into n adaptive gradient levels, where n ranges from 3 to 6. The loading step size of each level is the product of the target clamping force divided by the number of levels and the corresponding gradient weight coefficient. The gradient weight coefficient is adaptively adjusted as the number of loading levels increases. The loading step size of the first two levels does not exceed 20% of the target force, and the loading step size of the last level does not exceed 10% of the target force. After each loading level is completed, the load is held for 0.1s to 0.3s. During the holding phase, multi-dimensional force sensor data is collected in real time at a sampling frequency of 1kHz. A closed-loop control is formed through a fuzzy PID controller to ensure smooth force loading without overshoot. Plastic Deformation Precursor Sliding Mode Observation and Early Warning: A detonator plastic deformation precursor identification model based on a super-helical sliding mode observer is constructed. This model monitors the changes in the equivalent stiffness of the detonator shell and the slope of the force-displacement curve in real time during clamping. The observer employs a two-level state update rule: The first level calculates the update rate of the clamping force observation based on the deviation between the actual clamping force sample value and the observed clamping force value, combined with the first sliding mode gain, and synchronously updates the observed clamping force change rate. The second level updates the change rate of the observed clamping force change based on the sign function of the clamping force deviation and the second sliding mode gain. Preset instantaneous peak thresholds and yield precursor thresholds are used. When the instantaneous peak of the force feedback exceeds the instantaneous peak threshold, the system determines it as an impact or positioning error, immediately stops loading, and reverses unloading. When the observed decrease in the equivalent stiffness change rate exceeds the preset threshold, or the sudden change in the slope of the force-displacement curve exceeds the yield precursor threshold, the system determines that the detonator shell has entered the plastic deformation precursor stage, immediately stops loading, and triggers an audible and visual alarm. Clamping reliability verification: After the clamping force stabilizes within ±3% of the target clamping force, the system applies a small swing disturbance of preset amplitude to the end of the robotic arm, collects force feedback data in real time during the disturbance process, calculates the actual static friction coefficient online, and confirms the clamping reliability when the actual static friction coefficient is greater than the preset minimum safe friction coefficient, and enters the handling program.

[0013] Furthermore, in step S3, the mass, damping, and stiffness parameters of the robotic arm end are set so that when it encounters slight resistance, the robotic arm can passively adapt to avoid damage to the detonator or equipment due to rigid collision.

[0014] Furthermore, in step S4: Visual guidance entry: The visual recognition unit acquires images of the explosive loading port, identifies and calculates the central axis and spatial pose of the loading port, and the robotic arm guides the detonator tip to a preset depth in the loading port in a mode of position control as the main control and force supervision as the auxiliary control. At this stage, a safety upper limit for contact force is set. When the axial force feedback value exceeds the upper limit, the system immediately stops the movement and adjusts the detonator posture until the contact force meets the safety requirements. Online working condition identification and force control mode switching: The distance between the detonator tip and the cartridge / hole wall is detected in real time by the proximity sensor. Combined with the axial force feedback data of the multi-dimensional force sensor, when the proximity sensor detects that the detonator tip has entered the preset depth of the filling port, or the axial force feedback value continuously exceeds the preset filling start force threshold, the system identifies working condition parameters such as the equivalent friction coefficient and the equivalent stiffness of the cartridge in the filling port online based on the recursive least squares method. At the same time, it switches from the position-dominated mode to the variable impedance model prediction force control mode. Variable impedance model predicts compliant propulsion: A variable impedance model predictive controller based on a nonlinear disturbance observer is constructed to achieve high-precision tracking and adaptive adjustment of the loading force; First, an impedance mechanics model for the loading process at the end of the robotic arm is established. This model is based on inertial parameters, damping parameters, and stiffness parameters. The difference between the actual contact force and the target loading force is balanced by the weighted sum of the second and first derivatives of the position deviation and the position deviation. Based on the online identified working condition parameters, the system adaptively adjusts the damping and stiffness parameters in real time. When an increase in frictional resistance or a jamming trend is detected, the system automatically reduces the stiffness parameter and increases the damping parameter to improve the system's compliance. Secondly, a constrained model predictive control optimization objective function is constructed. The optimization objective includes two core components: the first part is the weighted sum of squares of the tracking errors between the predicted target loading force and the actual contact force in the prediction time domain, which is used to ensure the tracking accuracy of the loading force; the second part is the weighted sum of squares of the increments of the feed rate control quantity in the control time domain, which is used to ensure the stability of the control process. The optimization process needs to satisfy three types of constraints: the upper limit of the loading force safety limit, the upper limit of the feed rate, and the upper limit of the feed stroke. The optimal feed rate control sequence is obtained by solving in each control cycle to achieve zero steady-state error and high dynamic tracking of the loading force. Simultaneously, a nonlinear disturbance observer is constructed, and intermediate variables and observer gain are set. The intermediate variables are updated in real time with the change rate of position deviation, damping parameters, stiffness parameters, actual contact force, and feed rate control quantity during the filling process as inputs. Based on the intermediate variables, observer gain, and change rate of position deviation, the real-time observed values ​​of tunnel vibration, borehole wall friction fluctuation, and lumped disturbance caused by model uncertainty are calculated. The system feeds forward this disturbance observed value to the output of the model predictive controller to offset the influence of external disturbance on the filling force control. Intelligent identification and adaptive processing of multi-source abnormal working conditions: Based on multi-source data from multi-dimensional force sensors, displacement sensors, and proximity sensors, the system uses preset feature threshold rules to identify three types of abnormal working conditions in real time: ① Force overload condition: The force feedback value instantaneously exceeds the safe upper limit of the loading force; ② Jamming condition: The rate of change of axial displacement under continuous feed control is less than 0.1 mm / s and lasts for more than 0.5 s; ③ Unstable condition: The standard deviation of the force feedback value exceeds the preset fluctuation threshold. For different abnormal working conditions, the system executes differentiated adaptive processing strategies: For the force overload condition, the feed is immediately stopped and the system reverses and retreats 3 mm to 5 mm; for the jamming condition, the feed is stopped and the system reverses and retreats 2 mm to 3 mm, then the gripper is adaptively rotated to adjust its posture within ±15°, while the impedance parameter is adjusted to reduce stiffness, and the system attempts to feed again; for the unstable condition, the weight parameters of the model predictive controller are optimized in real time to reduce the feed speed and improve control stability; if the abnormality cannot be eliminated after 3 consecutive adaptive processing steps, the system triggers an audible and visual alarm and awaits manual intervention. Multi-feature fusion-based arrival judgment: The system integrates feed distance data from displacement sensors, force feedback data from multi-dimensional force sensors, and stable data from the time dimension to construct a multi-feature fusion-based arrival judgment model. When the following conditions are met simultaneously, the detonator is judged to be in place: ① The feed distance reaches the preset target filling depth; ② The axial force feedback value is stable near the target filling force, with a fluctuation range of less than ±2N; ③ There are no significant abrupt changes in displacement and force feedback values ​​within 1.5s; After the arrival judgment is made, the system enters the compliant release stage. Compliant release: The gripper uses a gradient reduction strategy to perform compliant release. Based on the target gripping force, the gripping force is gradually reduced in 3 to 5 levels, with each level held for 0.2 seconds. The unloading step of each level does not exceed 30% of the target gripping force. After the gripping force is completely unloaded and the gripper is fully opened, the robotic arm slowly exits the loading port at a preset low and uniform speed to avoid mechanical vibration from carrying out the detonator or causing impact.

[0015] Furthermore, the abnormal working condition adaptive handling strategy performs differentiated processing for three types of working conditions: force overload, jamming, and instability. The basic processing flow includes stopping the feed, slightly retracting by 2mm to 5mm, adjusting the posture of the adaptive rotating gripper within ±15°, optimizing the control parameters, and then attempting to load again. If the processing fails three times in a row, an audible and visual alarm is issued, and manual intervention is required.

[0016] Furthermore, in step S5, key parameters such as force, displacement, and speed are recorded during each clamping and loading process to establish an operation database. Through offline analysis, parameters such as target clamping force and loading force can be optimized to achieve self-learning capability based on historical data, thereby gradually enhancing the system's adaptability to different batches and working conditions.

[0017] The beneficial effects of this invention are as follows: 1. A fuzzy PID adaptive gradient clamping force control algorithm based on online stiffness identification is adopted. The stiffness and yield critical force of the detonator shell are identified online by recursive least squares method. The target clamping force and variable gradient loading strategy are adaptively generated. Combined with a dual-loop coupled fuzzy PID controller, the clamping force is controlled with no overshoot and precise control. This breaks through the limitations of traditional fixed gradient and fixed parameter PID control. The clamping parameters can be dynamically matched according to the shell characteristics of detonators of different specifications. This avoids the detonator being crushed by excessive clamping force, causing an explosion due to squeezing of the agent, or the detonator slipping due to insufficient clamping force. At the same time, the early identification and warning of the plastic deformation of the detonator is realized through the super-helical sliding mode observer. Combined with a high-friction flexible buffer layer, the clamping stability is improved while achieving safe, reliable and damage-free clamping throughout the process. The safety protection level and the adaptability of working conditions are greatly improved. 2. An interference observation-based variable impedance model prediction algorithm for compliant loading force control is adopted. By online identification of loading conditions and adaptive adjustment of impedance parameters, combined with constrained model predictive control, high-precision dynamic tracking of loading force is achieved. At the same time, a nonlinear interference observer is used to compensate for external interferences such as tunnel vibration and friction fluctuations in real time, which completely solves the problems of weak anti-interference ability, low tracking accuracy and insufficient compliance of traditional PI+feedforward control. With the help of a multi-feature fusion judgment model, it can avoid damage to detonators and fuses and crushing of explosives due to overload of loading force, and ensure that detonators and explosives are in close contact, which significantly improves the detonation success rate and blasting construction quality. 3. The detonator position recognition, safe clamping, compliant handling, precise loading, and compliant release are integrated into a unified control system to achieve seamless connection between each link. During the handling stage, impedance control is used to give the robotic arm compliant characteristics, which can passively adapt to external collisions and effectively avoid damage to the detonator or equipment failure caused by rigid impacts. The overall operation accuracy, continuity and stability are greatly improved. 4. Through multi-sensor data fusion and online identification algorithms, the dust resistance, vibration resistance and interference resistance of detonator clamping and loading are significantly enhanced. It maintains high-precision detection and control even in complex tunnel construction environments. It can adapt to different detonator models, different explosive types, and different hole diameters and depths. It does not require frequent manual parameter adjustments, and its versatility and operational efficiency are significantly improved. 5. Multiple safety redundancy mechanisms are set up, including instantaneous peak protection, early warning of plastic deformation slip mode observation, hardware-level safety circuit, and adaptive handling of multi-source abnormal working conditions. It can predict risks in real time, respond to faults quickly, and actively cut off dangerous outputs to curb safety accidents from the source. At the same time, it has the ability to record data throughout the process and learn offline. It can automatically iterate and optimize fuzzy rules, model prediction control parameters, impedance parameters, etc., and continuously improve the system's adaptability and operational stability to different detonator batches and different construction conditions. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the detonator safety clamping and loading force control system of the tunnel blasting charging robot of the present invention; Figure 2 This is a schematic diagram of the assembly structure of the detonator clamping and loading execution unit and the robotic arm of the present invention; Figure 3 This is a schematic diagram of the detonator clamping and loading execution unit of the present invention; Figure 4 This is a flowchart illustrating the entire process of the tunnel blasting charging robot's detonator safety clamping and loading force control method of the present invention. Figure 5 This is a block diagram of the multi-stage force-controlled safety clamping logic for detonators according to the present invention; Figure 6 This is a block diagram of the detonator compliant loading force control logic of the present invention; Figure 7 A comparison chart of the operational performance of conventional drilling and blasting operations in mountain tunnels; Figure 8 Comparison chart of the results of adaptive clamping force control compatibility test for multi-specification detonators; Figure 9 This is a comparison chart showing the anti-interference performance of the system under harsh working conditions of high dust and strong vibration. In the diagram, 1 is the robotic arm; 2 is the actuator quick-change interface; 3 is the gripper; 301 is the buffer layer; 4 is the vision recognition component; 401 is the high-definition industrial camera; 402 is the supplementary light source; and 5 is the proximity sensor. Detailed Implementation

[0019] The following embodiments illustrate the present invention in detail. In the description of these embodiments, specific details such as particular system structures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods are omitted so as not to obscure the description of this application with unnecessary detail.

[0020] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0021] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0022] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0023] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0024] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized. Example 1

[0025] Please see the appendix Figure 1-3 A tunnel blasting explosive loading robot detonator safety clamping and loading force control system includes a robot body, and a detonator clamping and loading execution unit and a control unit set on the robot body. The robot adopts a six-axis articulated robotic arm with six degrees of freedom of spatial motion capability, which can realize the grasping, transfer and loading of detonators in complex postures in the tunnel. The end of the robotic arm is equipped with an actuator quick-change interface. The interface adopts an ISO standard quick-change disk structure, which can realize the quick disassembly and positioning of the detonator clamping and loading execution unit. The repeatability of positioning is no more than ±0.05mm. The detonator clamping and loading execution unit is fixed to the flange face at the end of the six-axis robotic arm through the actuator quick-change interface and is arranged coaxially. It includes a clamping mechanism, a multi-dimensional force / torque sensor, a displacement sensor, a vision recognition component, and a proximity sensor. The clamping mechanism uses servo electric grippers, with the SMCLEHZ series electric grippers or OnRobotRG2-FT flexible grippers as the models. A high-friction coefficient, low-hardness flexible buffer layer is pasted on the inner side of the grippers. The buffer layer material is polyurethane (PU) or food-grade silicone, with a thickness of 2mm to 4mm. It is used to increase the clamping friction, buffer the impact, and protect the detonator shell from being pinched. The grippers can be fitted with multiple sets of finger tips to adapt to detonators of different diameters. The multi-dimensional force / torque sensor uses the ATINano17 six-dimensional force sensor or the RobotiqFT300-S torque sensor, which is rigidly installed between the electric gripper and the quick-change interface at the end of the robotic arm. It is located on the force transmission path between the end of the robotic arm and the gripping mechanism, and collects the three-dimensional axial forces Fx, Fy, Fz and three-dimensional torques Mx, My, Mz in real time during the gripping, handling and loading process. The sampling rate is not less than 1kHz, and the force signals are sent to the control unit in real time. The displacement sensor uses Keyence GT2-H12 high-precision contact displacement sensor, or Tamagawa absolute encoder integrated into the joint of the robotic arm or inside the gripper motor. The contact displacement sensor is installed inside the gripping mechanism, at the base of the gripper fingers, or the encoder built into the servo motor is used as displacement feedback and installed at the tail of the electric gripper drive motor. The measurement resolution is no higher than 0.1mm, and it is used to detect the gripper opening degree, detonator radial dimension, detonator axial feed displacement and spatial position in real time. The visual recognition component consists of at least one high-definition industrial camera and a supplementary light source. The industrial camera is a Hikvision MV-CA013-10GC 1.3-megapixel global shutter camera with an 8mm fixed-focus industrial lens. The supplementary light source is a 24V white ring LED light source, which is fixed to the side of the clamping mechanism and above the front end of the gripper via an L-shaped bracket. The lens faces the detonator tray and the explosive loading port to ensure that the field of view covers the detonator gripping position and the loading port. The proximity sensor uses an Omron E2K-C25ME1 capacitive proximity switch or a Panasonic HG-C1050 laser displacement sensor, which is installed at the front end of the gripper, close to the output end of the detonator. The detection direction is along the axis of the detonator towards the explosive hole or cartridge. It detects the distance and contact state between the front end of the detonator and the cartridge or hole wall in a non-contact manner. When the detection distance is no more than 3mm to 5mm, a trigger signal is output as a key criterion for switching the force control mode during the loading stage. The control unit includes a host computer, a slave computer, and a safety protection module. The host computer is an Advantech UNO-2484G embedded industrial computer, responsible for path planning, force control algorithm calculation, visual image processing, human-machine interaction, and data recording. The slave computer is a Beckhoff CX9020 motion controller or a Googol GUS-400 motion controller, responsible for receiving instructions and controlling the servo drive of the robotic arm and gripping mechanism in real time with a 1ms cycle, forming a high-bandwidth force closed loop. The safety protection module is a hardware-level safety circuit independent of the main control system, including a comparator circuit, relays, emergency stop circuit, and audible and visual alarm, integrated in the electrical control box and connected in series in the servo drive power supply circuit. When the gripping force or loading force exceeds the preset hardware threshold, the drive power can be cut off immediately and an alarm can be triggered to achieve the highest priority safety protection and prevent the risk of detonator overload damage or explosion.

[0026] Reference Appendix Figure 4 A force control method based on a tunnel blasting explosive loading robot detonator safety clamping and loading force control system includes the following steps: S1. Visual Recognition and Path Planning: The explosive loading robot acquires images of the detonator tray area through a visual recognition unit, identifies the detonator target and calculates its six-degree-of-freedom pose, plans a collision-free gripping path, and moves the gripper to the gripping starting point. The visual recognition unit acquires images of the detonator tray area using a high-definition industrial camera, identifies the detonator target through a deep learning-based YOLO or MaskR-CNN network, and uses the PnP (Perspective-n-Point) algorithm to calculate the detonator's central axis, end face position, and six-degree-of-freedom spatial pose. Based on the recognition results and combined with the robotic arm's kinematic model, the control system plans a collision-free gripping path. Reference Appendix Figure 5 S2, Multi-level force control gripping: The robot's robotic arm and end gripper adopt a fuzzy PID adaptive gradient gripping force control algorithm based on online stiffness identification. It executes a progressive control strategy that includes coarse position positioning, online identification of detonator shell stiffness, force / position hybrid fuzzy PID adaptive adjustment, variable gradient incremental loading, early warning of plastic deformation precursor sliding mode observation, and gripping reliability verification to complete the safe gripping of the detonator. Coarse positioning: The robotic arm moves the gripper to the outside of the detonator and closes the gripper quickly according to the trapezoidal speed plan in the position control mode. When the flexible buffer layer of the gripper makes slight contact with the detonator shell and the displacement sensor shows a continuous change in the gripper opening value, the system records the current initial opening and initial contact force to complete the preliminary positioning. Online Detonator Shell Stiffness Identification: Based on force-displacement sampling data during the initial contact stage, the system uses a recursive least squares method to identify the equivalent shell stiffness of the current detonator online. With yield critical force The formula is as follows:

[0027] Formula parameter description: Let be the vector of parameters to be identified at time k. The recursive gain matrix is... The force sensor sample value at time k. Let k be the regression vector composed of the displacement sensor sampling values ​​at time k; Based on the identified equivalent stiffness and yield critical force, the system adaptively generates the target clamping force. With the graded loading strategy, the formula for the target clamping force is as follows:

[0028] Formula parameter description: For detonator quality, It is the acceleration due to gravity. This is the maximum expected acceleration of the robotic arm. For safety reasons, The static friction coefficient between the gripper and the detonator housing; Force / Position Hybrid Fuzzy PID Adaptive Adjustment: The system switches to a force / position hybrid control mode, constructing a dual-loop coupled fuzzy PID controller with the gripper opening degree as the position control loop and the gripping force as the force control loop; the gripping force deviation e and the deviation change rate ec are used as inputs to the fuzzy controller, and the proportional coefficient of the PID controller is self-tuned online according to preset fuzzy rules. Integral coefficient Differential coefficients Fuzzy rules are adopted The Mamdani inference model uses the centroid method for fuzzy resolution to achieve overshoot-free and fast-response control of the clamping force; Variable gradient incremental loading: Detonator shell yield critical force based on online identification , to clamp the target It is divided into n adaptive gradient levels, where n ranges from 3 to 6, and the formula for the loading step size of each level is as follows:

[0029] Formula parameter description: The gradient weight coefficients are adaptively adjusted as the loading level increases. The loading step size of the first two levels does not exceed 20% of the target force, and the loading step size of the last level does not exceed 10% of the target force. After each loading level is completed, the load is held for 0.1s to 0.3s. During the holding phase, multi-dimensional force sensor data is collected in real time at a sampling frequency of 1kHz. A closed-loop control is formed through a fuzzy PID controller to ensure smooth force loading without overshoot. Early warning of plastic deformation precursors: A model for identifying precursors of plastic deformation in detonators based on a super-helical sliding mode observer is constructed. The changes in the equivalent stiffness rate and the slope of the force-displacement curve of the detonator shell during clamping are observed in real time. The state equation of the sliding mode observer is as follows:

[0030] Formula parameter description: This is the actual clamping force sample value. For the clamping force observation value, The observed value of the rate of change of clamping force. , For sliding mode gain; The system has preset instantaneous peak threshold and yield precursor threshold. When the instantaneous peak of the force feedback exceeds the instantaneous peak threshold, the system determines that it is an impact or positioning error, and immediately stops loading and unloads in the reverse direction. When the observed decrease in the equivalent stiffness change rate exceeds the preset threshold, or the sudden change in the slope of the force-displacement curve exceeds the yield precursor threshold, the system determines that the detonator shell has entered the plastic deformation precursor stage, immediately stops loading and triggers an audible and visual alarm. Clamping reliability verification: When the clamping force stabilizes at the target clamping force After the error range of ±3%, the system applies a small swing disturbance of preset amplitude to the end of the robotic arm, collects force feedback data in real time during the disturbance process, calculates the actual static friction coefficient online, and confirms that the clamping is reliable when the actual static friction coefficient is greater than the preset minimum safe friction coefficient, and enters the handling program. S3. Compliant handling: After the robotic arm grips the detonator, it moves to the loading position of the blast hole or explosive cartridge according to the planned path. The handling process adopts an impedance control strategy to plan the handling path and sets the mass, damping and stiffness parameters of the end of the robotic arm so that it exhibits compliance with external collisions during the movement. When encountering slight resistance, the robotic arm can passively comply to avoid damage to the detonator or equipment due to rigid collisions. Reference Appendix Figure 6 S4, Compliant Loading: When loading detonators, the gripper at the end of the robotic arm employs a compliant loading force control algorithm based on a variable impedance model predicted by interference observation. This algorithm executes a progressive process including visual guidance into the hole, online identification of working conditions and switching of force control modes, compliant advancement predicted by the variable impedance model, intelligent identification and adaptive processing of multi-source abnormal working conditions, multi-feature fusion judgment of placement, and gradient compliant release, loading the detonator to the target position. Specifically: Visual guidance entry: The visual recognition unit acquires images of the explosive loading port, identifies and calculates the central axis and spatial pose of the loading port, and the robotic arm guides the detonator tip to a preset depth in the loading port in a mode of position control as the main control and force supervision as the auxiliary control. At this stage, a safety upper limit for contact force is set. When the axial force feedback value exceeds the upper limit, the system immediately stops the movement and adjusts the detonator posture until the contact force meets the safety requirements. Online working condition identification and force control mode switching: The distance between the detonator tip and the cartridge / hole wall is detected in real time by the proximity sensor. Combined with the axial force feedback data of the multi-dimensional force sensor, when the proximity sensor detects that the detonator tip has entered the preset depth of the filling port, or the axial force feedback value continuously exceeds the preset filling start force threshold, the system identifies working condition parameters such as the equivalent friction coefficient and the equivalent stiffness of the cartridge in the filling port online based on the recursive least squares method. At the same time, it switches from the position-dominated mode to the variable impedance model prediction force control mode. Variable impedance model predicts compliant propulsion: A variable impedance model predictive controller based on a nonlinear disturbance observer is constructed to achieve high-precision tracking and adaptive adjustment of the loading force; First, an impedance model for the loading process at the end effector of the robotic arm is established, and the impedance parameters are adaptively adjusted, as shown in the following formula:

[0031] Formula parameter description: The inertia matrix, Here is the damping matrix. Here is the stiffness matrix. For positional deviation, This is the actual contact force feedback value. The system aims to achieve the target loading force; based on online identified operating parameters, it adaptively adjusts the loading force in real time. and When an increase in frictional resistance or a jamming trend is detected, the stiffness matrix is ​​automatically reduced. Increase the damping matrix

[0032] Improve system compliance; Secondly, the constrained model predictive control optimization objective function is constructed, as shown in the following formula:

[0033] in, To predict the time domain, To control the time domain, For the force tracking error weight matrix, For the control quantity increment weight matrix, for The control value of the end-effector feed speed at any given moment. To control the increment of the loading force, the objective function is optimized to satisfy the safety constraints of the loading force, the feed rate constraints, and the position and travel constraints. A quadratic programming problem is solved in each control cycle to obtain the optimal feed rate control sequence, achieving zero steady-state error and high dynamic tracking of the loading force. At the same time, a nonlinear disturbance observer is constructed to observe and compensate for lumped disturbances caused by tunnel vibration, borehole wall friction fluctuations, and model uncertainties in real time. The observer equation is:

[0034] in, For observer intermediate variables, For observer gain, The system feeds forward the disturbance observations to the output of the model predictive controller to compensate for the external disturbances and counteract their impact on the loading force control. Intelligent identification and adaptive processing of multi-source abnormal working conditions: Based on multi-source data from multi-dimensional force sensors, displacement sensors, and proximity sensors, the system uses preset feature threshold rules to identify three types of abnormal working conditions in real time: ① Force overload condition: The force feedback value instantaneously exceeds the safe upper limit of the loading force; ② Jamming condition: The rate of change of axial displacement under continuous feed control is less than 0.1 mm / s and lasts for more than 0.5 s; ③ Unstable condition: The standard deviation of the force feedback value exceeds the preset fluctuation threshold. For different abnormal working conditions, the system executes differentiated adaptive processing strategies: For force overload conditions, feed is immediately stopped and the system reverses and retreats 3 mm to 5 mm; for jamming conditions, feed is stopped and the system reverses and retreats 2 mm to 3 mm, then the gripper is adaptively rotated to adjust its posture within ±15°, while the impedance parameter is adjusted to reduce stiffness, and feed is attempted again; for unstable conditions, the weight parameters of the model predictive controller are optimized in real time to reduce the feed speed and improve control stability; if the abnormality cannot be eliminated after 3 consecutive adaptive processing steps, the system triggers an audible and visual alarm and awaits manual intervention. Multi-feature fusion-based arrival judgment: The system integrates feed distance data from displacement sensors, force feedback data from multi-dimensional force sensors, and stable data from the time dimension to construct a multi-feature fusion-based arrival judgment model. When the following conditions are met simultaneously, the detonator is judged to be in place: ① The feed distance reaches the preset target filling depth; ② The axial force feedback value is stable near the target filling force, with a fluctuation range of less than ±2N; ③ There are no significant abrupt changes in displacement and force feedback values ​​within 1.5s; After the arrival judgment is made, the system enters the compliant release stage. Compliant release: The gripper uses a gradient reduction strategy to perform compliant release. Based on the target gripping force, the gripping force is gradually reduced in 3 to 5 levels, with each level held for 0.2 seconds. The unloading step of each level does not exceed 30% of the target gripping force. After the gripping force is completely unloaded and the gripper is fully opened, the robotic arm slowly exits the loading port at a preset low and uniform speed to avoid mechanical vibration from carrying out the detonator or causing impact. S5. Data Recording and Self-Learning: By recording key parameters such as force, displacement, and speed during each clamping and loading process and establishing an operation database, the system can optimize parameters such as target clamping force and loading force through offline analysis, thereby achieving self-learning capability based on historical data and gradually enhancing the system's adaptability to different batches and working conditions.

[0035] In summary, the tunnel blasting charging robot detonator safety clamping and loading force control method and system disclosed in this invention addresses the urgent need for safer, more precise, and intelligent detonator loading operations in drill-and-blast tunnel construction. It addresses the common technical pain points of existing technologies, such as insufficient adaptive adjustment capability of clamping force, low loading force control accuracy, lack of coordination between clamping and loading, weak anti-interference performance in complex construction environments, and lack of safety redundancy and fault tolerance mechanisms. Through innovative construction of a multi-level gradient force control clamping, force / position hybrid compliant loading, multi-sensor information fusion, full-process collaborative control, hardware-level safety protection, and data-driven self-learning integrated technical solution, this invention provides a comprehensive solution. This invention employs a clamping control mode combining multi-level gradient increasing force loading with dynamic judgment of dual safety thresholds. Combined with a flexible buffer structure and real-time force feedback closed-loop adjustment, it can dynamically match the clamping force according to the detonator model, shell strength, and stress state, effectively avoiding major safety hazards such as detonator shell damage and explosive compression due to excessive clamping force, or detonator detachment due to insufficient clamping force. It also ensures a smooth, impact-free clamping process and reliable, non-deviation-prone positioning. Through force / position hybrid control, a compliant loading strategy using a PI controller and velocity feedforward, and the fusion judgment of multi-source information including displacement, force feedback, and time, it achieves high-precision adaptive tracking of loading force and accurate identification of detonator positioning. This prevents overload of loading force from damaging the detonator fuse and compressing the explosive, while ensuring a tight fit between the detonator and the explosive, significantly improving detonation reliability and blasting effect. Simultaneously, it integrates visual recognition, safe clamping, compliant handling, precise loading, and compliant release into a unified control system for coordinated operation. During the handling phase, an impedance control strategy imparts passive compliant characteristics to the robotic arm, significantly reducing the risk of rigid collisions. Leveraging a multi-sensor fusion design that integrates multi-dimensional force or torque sensors, precision displacement sensors, proximity sensors, and visual recognition units, the system significantly enhances its detection stability and adaptability in harsh tunnel environments such as dust, vibration, and temperature fluctuations. It is compatible with different specifications of detonators, different types of explosives, and borehole operations with varying diameters and depths, eliminating the need for frequent manual parameter adjustments. Furthermore, it incorporates multiple safety redundancy mechanisms, including instantaneous peak value protection, yield pre-yield prediction, hardware-level safety circuits, and automatic anomaly handling. These mechanisms enable real-time risk prediction, rapid fault response, and proactive isolation of dangerous outputs, preventing accidents at their source. Through full-process data recording and offline self-learning optimization, the system continuously iterates control parameters, constantly improving its adaptability and operational stability to different detonator batches and construction conditions. Therefore, it completely solves the inherent defects of traditional manual loading of detonators in tunnel blasting and existing automated equipment in terms of safety, stability, accuracy and versatility, greatly improves the safety level, operation efficiency and intelligence of tunnel loading construction, effectively reduces the safety risks caused by manual intervention, and has outstanding advantages such as high safety, strong adaptability, precise control, reliable operation and high intelligence. It can effectively meet the industrial needs of intelligent, unmanned and standardized construction of modern tunnel engineering. Example 2

[0036] The conventional drilling and blasting construction scenario in Class II surrounding rock of a mountain tunnel is the most commonly used basic working condition, and it is used to verify the core control performance of the algorithm of this patent.

[0037] 1. Hardware configuration and parameters of the controlled object Robot body: Six-axis articulated robotic arm, rated load 5kg, repeatability ±0.03mm, end effector quick-change interface repeatability no more than ±0.05mm; End effector: SMCLEHZ series servo electric gripper with a 3mm thick polyurethane flexible buffer layer bonded to the inside; ATINano17 six-dimensional force / torque sensor with a sampling rate of 1kHz; Keyence GT2-H12 high-precision displacement sensor with a resolution of 0.05mm; Hikvision 1.3-megapixel global shutter industrial camera with a ring LED fill light source; Omron capacitive proximity sensor with a detection distance of 0 to 10mm. Controlled objects: No. 8 industrial galvanized iron shell millisecond delay electric detonator, 8mm in diameter, 50mm in length, 12g in weight per piece, pre-calibrated value of shell yield critical force 120N; No. 2 rock emulsion explosive cartridge, 32mm in diameter, 42mm in borehole diameter, 3m in borehole depth. Control unit: Advantech UNO-2484G industrial PC + Beckhoff CX9020 motion controller, control cycle 1ms.

[0038] 2. Core control parameter settings Clamping algorithm parameters: safety factor Pre-calibrated value of static friction coefficient Maximum expected acceleration of the robotic arm The adaptive calculation value of the target clamping force is 1.63N; the gradient loading is divided into 4 levels with weight coefficients of 0.2, 0.3, 0.3, and 0.2, respectively, and each level is held for 0.2s; the input and output universe of discourse of the fuzzy PID controller is [-6,6], the fuzzy subset has 7 levels, and there are a total of 49 fuzzy rules; the sliding mode gain of the super-spiral sliding mode observer is k1=120, k2=1500, the instantaneous peak threshold is set to 96N (80% of the yield critical force), and the yield precursor threshold is set to a 40% decrease in the stiffness change rate; Loading algorithm parameters: target loading force calibration value 8N; impedance model initial parameters Model predictive controller prediction time domain Control Time Domain The weight matrix is ​​Q=100, R=0.1; the nonlinear disturbance observer gain is L=150; and the safe upper limit of the loading force is set to 15N.

[0039] 3. Implementation process and measured data This embodiment successfully completed 100 consecutive detonator clamping, handling, and loading operations. The core measured data are as follows: In the visual recognition stage: the average time for detonator pose recognition is 12ms, the pose calculation error is no greater than 0.2mm, and the accuracy of filling port recognition is 100%. Clamping control: The average time for online identification of detonator shell stiffness is 8ms, with an identification accuracy of no less than 94%; the maximum overshoot of clamping force during the entire loading process is no more than 2.8%, and the fluctuation range of clamping force after stabilization is no more than ±2.5%, with no instantaneous spike exceeding the standard; no plastic deformation or slippage of the detonator shell occurred throughout the process; During the handling process: Under impedance control, the robotic arm experiences no rigid collisions throughout the entire process, the maximum contact force is no greater than 2N, and the detonator's attitude deviation is no greater than 0.15°; Loading control: The average error of loading force tracking is no greater than ±3%, and the maximum dynamic error is no greater than 4.2%; there is no jamming or excessive friction in the borehole; the accuracy of the arrival judgment is 100%, and the arrival depth error is no greater than 0.3mm; Release process: Gradient smooth release with no detonator displacement throughout the entire process; after the robotic arm exits, the detonator position shift is no more than 0.2mm.

[0040] 4. Comparison of Implementation Results Reference Appendix Figure 7 Compared with existing fixed clamping force + fixed loading force technology, in this embodiment: the clamping force overshoot is reduced by 82%, the loading force tracking accuracy is improved by 70%, the detonator shell damage rate is reduced from 1.2% to 0, the average time per operation cycle is shortened from 18s to 12s, and the operation success rate is increased from 92% to 100%. Example 3

[0041] The existing technology cannot adapt to the pain point of different detonator specifications, thus verifying the multi-condition adaptive capability of the online identification algorithm of this patent.

[0042] 1. Test Objects and Parameter Settings Three commonly used detonators in engineering were tested without any manual modification of any control parameters. The system automatically identified and adaptively matched control strategies online. The specific detonator parameters are as follows:

[0043] Core algorithm parameters: recursive least squares forgetting factor 0.95, identification period 1ms; gradient hierarchical adaptive adjustment to 3 to 6 levels, the loading step size of the first two levels does not exceed 20% of the target force; the rest of the hardware configuration is the same as in Example 1.

[0044] 2. Actual Measurement Data and Implementation Results Reference Appendix Figure 8 Each type of detonator was continuously tested 50 times in the entire process. The core measured data are as follows: Stiffness identification performance: The online identification accuracy of the shell equivalent stiffness and yield critical force of the three types of detonators is not less than 92%, the maximum identification error is not greater than 7.5%, and the identification time is not greater than 10ms. Clamping control performance: No. 6 plastic shell detonator: adaptively generates target clamping force of 1.09N, clamping force overshoot is no more than 3.2%, stability error is no more than ±2.8%, and there are no shell deformation or breakage issues; No. 8 iron-cased detonator: Same as in Example 2; Aluminum alloy shell digital detonator: adaptively generates target clamping force of 2.04N, clamping force overshoot is no more than 2.5%, stability error is no more than ±2.2%, and there is no slippage throughout the entire process; Compatibility Comparison: When the existing technology uses a fixed clamping force parameter of 1.6N, the No. 6 plastic shell detonator shows a shell deformation rate of 3.5%, and the aluminum alloy digital detonator shows a slippage rate of 2%. Under the adaptive algorithm of this patent, the shell damage rate and slippage rate of the three types of detonators are all 0, the operation success rate is 100%, and no manual intervention is required to adjust the parameters. Example 4

[0045] The harsh construction environment of high dust and strong vibration at the tunnel face was simulated to verify the anti-interference ability and environmental adaptability of the patented algorithm.

[0046] 1. Operating condition simulation and parameter setting Interference environment: Background vibration generated by the synchronous operation of the rock drill, with a vibration frequency of 5Hz to 50Hz and an amplitude of ±0.2mm; a dust generator is used to simulate the high dust environment of the tunnel, with a dust concentration of 200mg / m³ (the conventional high dust concentration in tunnel construction). Hardware and algorithm parameters: consistent with Example 1, the nonlinear interference observer is turned on throughout the process, and the visual recognition adopts the anti-dust exposure compensation algorithm; the blast hole is set as an oblique blast hole with an inclination angle of 15° and a hole depth of 3.5m to simulate the non-horizontal hole working conditions on site.

[0047] 2. Actual Measurement Data and Implementation Results Reference Appendix Figure 9 The system successfully completed 80 full-process operations consecutively, and the core measured data are as follows: Vibration interference resistance: The nonlinear interference observer has a lumped interference compensation rate of no less than 91% for vibration-induced interference, and the force fluctuation caused by vibration during clamping is no greater than ±4%, with no overshoot or slippage; the filling force tracking error during filling is no greater than ±4.5%, with no jamming or impact caused by vibration; Dust interference resistance: In a high dust environment, the visual recognition unit has an accuracy rate of no less than 98% in recognizing detonators and filling ports, a pose calculation error of no more than 0.35mm, and no recognition failures. Comparative results: Under the same vibration environment, the existing technology has a clamping force fluctuation of more than 22%, three instances of detonator slippage, and a jamming rate of 12.5% ​​during the loading process; the success rate of this patented algorithm is 98.75%, with only one instance of manual intervention triggered by extreme dust obstruction, and no safety anomalies. Example 5

[0048] The effectiveness and safety of the abnormal handling mechanism of this patent are verified in response to common abnormal conditions in tunnel construction, such as borehole deviation, jamming, and force overload.

[0049] 1. Abnormal operating condition simulation and parameter setting Simulated anomaly types: ① Hole deviation and jamming condition: The hole axis is deviated by 3°, simulating the drilling deviation in the field; ② Force overload condition: The detonator positioning deviation is 2mm, resulting in sudden force exceeding the standard during the clamping / loading process; ③ Rough and unstable hole wall condition: Protruding obstacles are set on the inner wall of the hole to simulate the force fluctuation caused by uneven drilling. Hardware and algorithm parameters: consistent with Example 1, exception handling strategy is enabled throughout, maximum number of consecutive retries is 3, attitude adjustment range is ±15°.

[0050] 2. Actual Measurement Data and Implementation Results Each abnormal operating condition was simulated 50 times, and the core measured data are as follows: Handling jamming conditions: The system identifies jamming abnormalities within an average of 0.3 seconds and executes the process of "stop feeding - retreat 3mm - rotate 10° to adjust attitude - reduce stiffness by 50% and retry". The success rate of a single retry is 92%, and the overall success rate of three consecutive retryes is 100%. There are no cases of detonator damage or blast hole blockage. Force overload handling: When the instantaneous force exceeds the limit during the clamping process, the system triggers protection within an average of 0.1ms, immediately stops loading and unloads in the reverse direction, without plastic deformation of the detonator shell; when the force is overloaded during the filling process, the system immediately stops feeding and retreats 5mm, with an abnormality identification accuracy of 100%, and no damage to the detonator fuse or extrusion of explosives. Unstable working conditions: When force fluctuations are caused by rough hole walls, the system adjusts the impedance parameters and feed speed in real time. The filling process is free of impact and jamming, and the force fluctuation is controlled within ±5%. Comparative results: Under the same abnormal working conditions, the existing technology has an 8% detonator damage rate in the jammed working condition and a 2% safety accident trigger rate in the force overload working condition, and has no automatic attitude adjustment and retry capability; the algorithm of this patent has no detonator damage or safety abnormalities throughout the process, and the autonomous handling rate of abnormal working conditions is 98%. Example 6

[0051] This verifies the data recording and self-learning capabilities of this patent, demonstrating the system's long-term stability and iterative optimization effectiveness.

[0052] 1. Test Plan The system continuously completed 500 full-process operations, recording key parameters such as clamping force, loading force, displacement, and error for each operation. Offline parameter optimization was performed every 100 operations, iteratively updating the fuzzy PID control rules, model predictive control weights, and initial impedance parameters. The hardware and initial parameters were consistent with those in Example 1.

[0053] 2. Actual Measurement Data and Optimization Results 3. Implementation Results

[0054] Through 500 self-learning iterations, the system's clamping force control accuracy has improved by 34.6%, loading force tracking accuracy by 37.9%, and single operation time has been reduced by 10.7%. The adaptability to detonators of the same batch and blast holes under the same working conditions has been continuously improved. It has solved the defects of existing technology parameters being fixed and unable to be optimized during operation. The long-term operational stability is significantly better than existing technology.

[0055] The above-described 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 detonator safety clamping and loading force control system for a tunnel blasting explosive loading robot, comprising a robot body, a detonator clamping and loading execution unit and a control unit disposed on the robot body, characterized in that, The robot body includes a six-axis robotic arm, the end of which is detachably connected to a detonator clamping and loading execution unit via an actuator quick-change interface. The detonator clamping and loading execution unit includes a clamping mechanism, a multi-dimensional force / torque sensor, a displacement sensor, a vision recognition component, and a proximity sensor. The multi-dimensional force / torque sensor is rigidly mounted between the clamping mechanism and the end of the robotic arm. The displacement sensor has a resolution no higher than 0.1mm. The vision recognition component includes a high-definition industrial camera and a supplementary light source. The proximity sensor is mounted on the front end of the gripper. The control unit includes a host computer, a slave computer, and an independent hardware-level safety protection module. The host computer is used for path planning, force control algorithm calculation, visual image processing, human-machine interaction, and data recording. The slave computer receives instructions from the host computer and controls the servo drive of the robotic arm and clamping mechanism in real time with a 1ms cycle to form a high-bandwidth force closed loop. The safety protection module is a hardware-level safety circuit connected in series in the servo drive power supply circuit, used to cut off the drive power and trigger an alarm when the clamping force or loading force exceeds a preset hardware threshold.

2. The tunnel blasting explosive loading robot detonator safety clamping and loading force control system according to claim 1, characterized in that, The clamping mechanism includes a servo-electric gripper with a flexible buffer layer attached to the inner side of the gripper; the multi-dimensional force / torque sensor is a strain gauge or piezoelectric sensor used to collect axial force, radial force and torque in real time during clamping, handling and loading processes, with a sampling frequency of not less than 1kHz; the displacement sensor is installed inside the clamping mechanism, or an encoder built into the robotic arm joint is used as a displacement detection element; the proximity sensor is a laser rangefinder or a capacitive proximity switch used to detect the contact state between the detonator tip and the cartridge and hole wall, serving as an auxiliary criterion for force control mode switching.

3. The tunnel blasting explosive loading robot detonator safety clamping and loading force control system according to claim 1, characterized in that, The actuator quick-change interface adopts the ISO standard quick-change disk structure, and the repeatability of positioning is no more than ±0.05mm; the high-definition industrial camera and the supplementary light source of the vision recognition component are fixed to the side of the clamping mechanism by the bracket, with the lens facing the detonator gripping position and the explosive loading port.

4. A force control method for the safe clamping and loading of detonators in a tunnel blasting explosive loading robot, characterized in that, Includes the following steps: S1. Visual Recognition and Path Planning: The explosive loading robot collects images of the detonator tray area through the visual recognition unit, identifies the detonator target and calculates its six-degree-of-freedom pose, plans a collision-free clamping path, and moves the gripper to the clamping starting point. S2. Multi-level force control gripping: The robotic arm and the end effector gripper adopt a fuzzy PID adaptive gradient gripping force control algorithm based on online stiffness identification. It sequentially executes progressive control of coarse position positioning, online identification of detonator shell stiffness, force / position hybrid fuzzy PID adaptive adjustment, variable gradient incremental loading, early warning of plastic deformation precursor sliding mode observation, and gripping reliability verification to complete the safe gripping of the detonator. S3. Compliant handling: After the robotic arm grips the detonator, an impedance control strategy is used to plan the handling path and set the target inertia, damping, and stiffness parameters at the end of the robotic arm to make the robotic arm compliant with external collisions during the movement. S4. Compliant loading: The gripper at the end of the robotic arm adopts a variable impedance model prediction compliant loading force control algorithm based on interference observation. It sequentially executes the progressive process of visual guidance into the hole, online identification of working conditions and switching of force control mode, variable impedance model prediction compliant propulsion, intelligent identification and adaptive processing of multi-source abnormal working conditions, multi-feature fusion judgment of arrival and gradient compliant release to load the detonator into the target position. S5. Data Recording and Self-Learning: Record key parameters of force, displacement, and speed during each clamping and loading process, establish an operation database, and optimize control parameters through offline analysis.

5. The force control method for safe clamping and loading of detonators in a tunnel blasting charging robot according to claim 4, characterized in that, In step S1, the visual recognition unit acquires images of the detonator tray area using a high-definition industrial camera, identifies the detonator target using a deep learning-based YOLO or MaskR-CNN network, calculates the detonator's central axis, end face position, and six-degree-of-freedom spatial pose using a perspective n-point algorithm, and the control system plans a collision-free clamping path in conjunction with the robotic arm's kinematic model.

6. The force control method for safe clamping and loading of detonators in a tunnel blasting explosive loading robot according to claim 4, characterized in that, The specific process of step S2 is as follows: Coarse positioning: The robotic arm moves the gripper to the outside of the detonator and plans to close the gripper in a trapezoidal speed using the position control mode. When the flexible buffer layer of the gripper contacts the detonator shell and the displacement sensor shows a continuous change in the gripper opening value, the system records the current initial opening and initial contact force to complete the preliminary positioning. Online identification of detonator shell stiffness: Based on the force-displacement sampling data during the initial contact stage, the system uses the recursive least squares method to identify the equivalent stiffness and yield critical force of the current detonator shell online. Based on the identification results, the system adaptively generates the target clamping force and graded loading strategy. The target clamping force is the product of the detonator mass, gravitational acceleration, maximum expected acceleration of the robotic arm and safety factor, and then divided by the static friction coefficient between the gripper and the detonator shell. Force / position hybrid fuzzy PID adaptive adjustment: The system switches to force / position hybrid control mode, using the gripper opening degree as the position control loop and the gripping force as the force control loop to construct a dual-loop coupled fuzzy PID controller. The clamping force deviation and deviation change rate are taken as inputs, and the proportional coefficient, integral coefficient, and derivative coefficient of the PID controller are self-tuned online through preset fuzzy rules to achieve overshoot-free control of the clamping force. Variable gradient incremental loading: Based on the online identification of the detonator shell yield critical force, the target clamping force is divided into 3 to 6 adaptive gradient levels. The loading step size of the first two levels does not exceed 20% of the target clamping force, and the loading step size of the last level does not exceed 10% of the target clamping force. After each loading level is completed, it is held for 0.1s to 0.3s. During the holding phase, force data is collected in real time at a sampling frequency of 1kHz to form a closed loop control. Plastic Deformation Precursor Sliding Mode Observation and Early Warning: Construct a detonator plastic deformation precursor identification model based on a super-helical sliding mode observer, and observe the changes in the equivalent stiffness change rate and the slope of the force-displacement curve of the detonator shell in real time during the clamping process. Preset instantaneous peak threshold and yield precursor threshold. When the instantaneous peak of the force feedback exceeds the instantaneous peak threshold, the system immediately stops loading and unloads in the reverse direction. When the observed decrease in the equivalent stiffness change rate exceeds the preset threshold or the sudden change in the slope of the force-displacement curve exceeds the yield precursor threshold, the system stops loading and triggers an audible and visual alarm. Clamping reliability verification: After the clamping force stabilizes within ±3% of the target clamping force, the system applies a small swing disturbance of preset amplitude to the end of the robotic arm and calculates the actual static friction coefficient online. When the actual static friction coefficient is greater than the preset minimum safe friction coefficient, the clamping reliability is confirmed.

7. The force control method for safe clamping and loading of detonators in a tunnel blasting explosive loading robot according to claim 4, characterized in that, In step S3, the impedance parameters at the end of the robotic arm are adaptively adjusted according to the detonator specifications and handling conditions. When the contact force exceeds a preset threshold during movement, the robotic arm passively adjusts its posture to offset the impact of the collision.

8. The force control method for safe clamping and loading of detonators in a tunnel blasting explosive loading robot according to claim 4, characterized in that, The specific process of step S4 is as follows: Visual guidance entry: The visual recognition unit acquires images of the explosive loading port, identifies and calculates the central axis and spatial pose of the loading port, and the robotic arm guides the detonator tip to a preset depth in the loading port in a mode of position control as the main control and force supervision as the auxiliary control. At this stage, a safety upper limit for contact force is set. When the axial force feedback value exceeds the upper limit, the system pauses the movement and adjusts the detonator posture. Online working condition identification and force control mode switching: The distance between the detonator tip and the cartridge / hole wall is detected in real time by the proximity sensor. Combined with the axial force feedback data of the multi-dimensional force sensor, when the proximity sensor detects that the detonator tip has entered the preset depth of the filling port, or the axial force feedback value continuously exceeds the preset filling start force threshold, the system identifies the equivalent friction coefficient and equivalent stiffness of the cartridge in the filling port online based on the recursive least squares method. At the same time, it switches from the position-dominated mode to the variable impedance model prediction force control mode. Variable Impedance Model Predictive Compliant Propulsion: A variable impedance model predictive controller based on a nonlinear disturbance observer is constructed. First, an impedance mechanics model of the loading process at the end of the robotic arm is established, with inertial parameters, damping parameters, and stiffness parameters as the core. The difference between the actual contact force and the target loading force is balanced by the weighted sum of the second and first derivatives of the position deviation and the position deviation. The system adaptively adjusts the damping and stiffness parameters in real time based on the online identified working condition parameters. Second, a constrained model predictive control optimization objective function is constructed, and the optimal feed rate control sequence is obtained by solving it in each control cycle. At the same time, a nonlinear disturbance observer is constructed to observe the lumped disturbances in the loading process in real time and feedforward compensation to the controller output. Intelligent identification and adaptive processing of multi-source abnormal working conditions: The system identifies three types of abnormal working conditions in real time based on multi-source sensor data: ① Force overload condition: The force feedback value instantly exceeds the safe limit of the loading force; ② Jamming condition: The rate of change of axial displacement under continuous feed control is less than 0.1 mm / s and the duration exceeds 0.5 s; ③ Unstable condition: The standard deviation of the force feedback value exceeds the preset fluctuation threshold; Implement differentiated adaptive handling strategies for different abnormal operating conditions; Multi-feature fusion judgment: The system integrates displacement, force feedback and time stability data to construct a judgment model for the detonator filling. The detonator is judged to be in place when the following conditions are met: ① The feed distance reaches the preset target filling depth; ② The axial force feedback value is stable near the target filling force, and the fluctuation range is less than ±2N; ③ There is no sudden change in displacement and force feedback values ​​exceeding the preset threshold within 1.5s. Compliant release: The gripper uses a gradient reduction strategy to perform compliant release. Based on the target gripping force, the gripping force is gradually reduced in 3 to 5 levels, with each level held for 0.2s. The unloading step of each level does not exceed 30% of the target gripping force. After the gripping force is completely unloaded and the gripper is fully opened, the robotic arm exits the loading port at a uniform speed not exceeding 5mm / s.

9. The force control method for safe clamping and loading of detonators in a tunnel blasting explosive loading robot according to claim 8, characterized in that, In the aforementioned adaptive handling strategy for abnormal working conditions, for force overload conditions, the system immediately stops feeding and reverses 3mm to 5mm; for jamming conditions, the system stops feeding and reverses 2mm to 3mm, the adaptive rotating gripper adjusts its posture within a range of ±15°, and at the same time adjusts the impedance parameters to reduce stiffness before attempting feeding again; for unstable working conditions, the system optimizes the controller weight parameters in real time and reduces the feeding speed; if the abnormality cannot be eliminated after three consecutive adaptive processing steps, the system triggers an audible and visual alarm and awaits manual intervention.

10. The force control method for safe clamping and loading of detonators in a tunnel blasting explosive loading robot according to claim 4, characterized in that, In step S5, the system analyzes the work database offline, iteratively optimizes the fuzzy PID control rules, model predictive control weights, and initial values ​​of impedance parameters, and achieves self-learning optimization based on historical work data, thereby improving the system's adaptability to different detonator batches and different construction conditions.

Images