Bionic flexible tactile sensor based on pressure pain and robot active protection method

Abstract

The application discloses a kind of based on pressure pain sensation's bionic flexible touch sensor and robot active protection method, it is related to bionic robot flexible sensing and safety protection field.The sensor adopts DIW integrated printing process, including flexible silica gel base, cross intersection row and column conductive grid and spherical cap micro convex pressure resistance sensing unit, no layering, full flexibility, can adapt to robot complex curved surface whole body.It is based on the critical pressure intensity of sensor material destruction as benchmark, constructs three-stage progressive quantization bionic pain sensation system, through pain sensation value driving robot executes graded protection action, core through expanding force area to eliminate structural risk from root;While constructing whole body partition density bionic skin system, fusion foot bottom pressure and gyroscope data realizes walking balance adjustment.The application gives consideration to sensing accuracy and engineering cost, greatly improves the bionic sensing ability of humanoid robot, operation safety and structure life.

Description

Technical Field

[0001] This invention relates to the field of biomimetic robot flexible electronic sensing technology, specifically to a biomimetic flexible tactile sensor based on pressure pain sensation and a robot active protection method. Background Technology

[0002] With the rapid development of humanoid robot technology, bionic tactile perception has become a core capability for robots to achieve precise manipulation, human-robot interaction, and autonomous safety protection. However, existing technologies suffer from the following key deficiencies, hindering the large-scale application of humanoid robots:

[0003] 1. Traditional resistive touch solutions use a two-layer rigid ITO conductive layer separation structure, which can only realize on / off point detection and cannot continuously measure pressure. In addition, the structure is rigid and easy to delaminate, which cannot be adapted to the curved surface of robot fingertips and repeated bending scenarios, and is difficult to apply to biomimetic flexible tactile perception.

[0004] 2. Existing flexible tactile sensors mostly adopt screen printing, multi-layer stacking or post-assembly processes, which have problems such as easy breakage of wires, poor surface fit, short bending life and high manufacturing cost, and cannot meet the full-area tactile perception needs of humanoid robots with complex curved surfaces.

[0005] 3. Currently, robot tactile perception can only achieve basic pressure and position detection. It has not established a quantitative pain mechanism based on the critical pressure of sensor material structure damage, nor does it have a graded risk warning system. It cannot replicate the progressive physiological protection logic of humans: "no discomfort from slight contact - pain from approaching damage - severe pain from damage".

[0006] 4. Existing robot safety protection solutions are mostly passive emergency actions such as shutdown and emergency retraction, lacking active protection capabilities. They cannot eliminate the risk of structural damage from the physical source, nor can they achieve a complete closed loop of "early warning - active protection - risk elimination - resumption of operation", which seriously affects the continuity of robot operation and structural life.

[0007] 5. Existing robotic bionic skins mostly adopt a uniform density layout without referring to the distribution pattern of human skin receptors for zonal design. This presents a core contradiction: "insufficient precision in the fine operation area" and "excessive cost and computational power consumption of the high-density layout throughout the body." Furthermore, there is no specific material and functional adaptation for foot-bearing scenarios, and gyroscope attitude data is not integrated to achieve dynamic walking balance adjustment, making it impossible to meet the dual needs of fine hand operation and lower limb walking balance. Summary of the Invention

[0008] 1. Technical problems to be solved

[0009] The purpose of this invention is to overcome the above-mentioned defects of the prior art and provide a biomimetic flexible tactile sensor based on pressure pain sensation and an active protection method for robots. It not only solves the structural defects of existing flexible sensors, such as easy delamination, easy breakage, and inability to adapt to complex curved surfaces, but also establishes a three-level progressive quantitative biomimetic pain sensation system, realizing graded active closed-loop protection for robots. At the same time, it constructs a zoned density biomimetic skin system that can cover the entire body of the robot, integrates multi-sensor data to achieve dynamic walking balance adjustment, and takes into account perception accuracy, engineering cost and operational safety.

[0010] 2. Technical Solution

[0011] To achieve the above objectives, the present invention adopts the following technical solution:

[0012] On the one hand, this invention provides an active protection method for robots, the core of which is a three-level progressive robot pain perception quantification system, the specific steps of which are as follows:

[0013] S1. Using the biomimetic flexible tactile sensor mounted on the robot, the force data of each sensing unit is collected in real time. The real-time pressure of a single point or area is calculated based on the force per unit area, and the robot's pain value is calculated based on the real-time pressure. The robot's pain value is a quantitative technical parameter that characterizes the risk of damage to the robot's body structure. It has no subjective physiological description, and all values ​​are calculated through objective physical parameters. It is reproducible and calibrable.

[0014] S2. Based on the structural characteristics of the robot, a three-level progressive threshold is set in advance for the robot's pain sensitivity value, namely the pre-trigger threshold, the trigger threshold, and the damage value, which correspond to different structural damage risk levels.

[0015] S3. Based on the real-time calculation of the robot's pain perception value and the matching relationship with the corresponding threshold, determine the structural damage risk level of the corresponding part, and execute the corresponding graded protection actions to achieve full-link protection of "early warning - active protection - emergency loss prevention".

[0016] Furthermore, the method for calculating the robot's pain value is as follows: the critical pressure BP for irreversible damage to the sensor and the corresponding body structure is pre-calibrated, the pressure P=F / S is calculated in real time, and the robot's pain value Pain=(P / BP)×100 is calculated. The pain value is positively correlated with pressure and damage risk.

[0017] Furthermore, the three threshold levels are based on BP calibration and correspond to three levels: potential risk, impending damage, and already damaged, with different threshold coefficients set for different parts of the organism.

[0018] Furthermore, the pain trigger point is accurately located by using a row and column matrix time-division scanning method, with no crosstalk and no ghost points, and the positioning accuracy can reach ±0.05mm.

[0019] Furthermore, for foot scenarios, the device integrates posture data from the gyroscope to establish a dual-level calibration mechanism for both feet and one foot, enabling precise balance adjustment in dynamic walking / running scenarios and deeply coupling it with the pain protection system.

[0020] On the other hand, this invention provides a biomimetic flexible tactile sensor based on pressure-induced pain sensation, serving as the core hardware carrier for the aforementioned active protection method. It is fabricated using a direct-write (DIW) integrated printing process, resulting in a fully flexible structure without layers or splicing. This structure includes a flexible silicone substrate, a cross-shaped conductive grid, and spherical micro-convex piezoresistive sensing units. The flexible silicone substrate can adapt to the complex curved surfaces of a robot's entire body. The cross-shaped conductive grid is embedded within the substrate, and the micro-convex piezoresistive sensing units are positioned at the intersections of the rows and columns, enabling force-to-electricity conversion and precise pressure detection.

[0021] 3. Beneficial effects

[0022] Compared with the prior art, the present invention has the following outstanding advantages:

[0023] 1. Structural level: The sensor of this invention adopts the DIW integrated printing process, which realizes the integrated molding of substrate, conductive grid and sensing unit, completely solving the defects of traditional sensors such as delamination, easy breakage of wires and inability to adapt to complex curved surfaces. It is resistant to bending and impact, and the bending life is more than 10 times that of existing multi-layer stacked sensors.

[0024] 2. Perception level: This invention pioneers a three-level progressive quantitative bionic pain sensing system. Based on the critical pressure of material damage, it completely replicates the progressive physiological protection logic of humans, solving the core pain point of existing single-threshold technologies such as "untimely warning and high false trigger rate", and realizing full-cycle monitoring from potential risks to irreversible damage.

[0025] 3. Intelligent level: This invention is the first to realize a complete closed-loop protection of "pain perception classification - graded protection - active risk elimination", which breaks through the existing protection mode of passive shutdown of robots. By expanding the force-bearing area, the pressure is reduced from the physical source, enabling the robot to have biological-like active self-rescue ability, which greatly improves the continuity of operation and structural safety.

[0026] 4. Full-domain coverage: This invention constructs a full-body bionic skin system for robots. By referring to the sensitivity rules of human skin, it conducts a zoned density layout and differentiated materials and threshold adaptation, which perfectly solves the core contradiction of "precision and cost". It can simultaneously cover the full-scenario needs of fine hand operation, foot walking balance and torso safety protection.

[0027] 5. Balance control level: This invention integrates plantar pressure distribution data and body gyroscope attitude data to establish a dual-level calibration mechanism for bipedal / single-pedal movement, achieving precise balance adjustment in dynamic walking / running scenarios. Unlike the pre-programmed gait of existing industrial robots, it has extremely strong terrain adaptability and is deeply coupled with the pain protection system.

[0028] 6. Engineering aspects: The sensor fabrication process of this invention is simple and can be mass-produced. The hardware only requires a matrix scanning circuit + a single ADC to achieve high-density point detection, without the need for a large number of chips and complex wiring. The hardware cost is reduced by more than 90% compared with existing MEMS tactile sensors, making it extremely suitable for large-scale applications in humanoid robots. Attached Figure Description

[0029] Figure 1 is a schematic diagram of the cross-grid structure of the biomimetic flexible tactile sensor of the present invention: 1 - row conductive lines; 2 - column conductive lines; 3 - micro-convex piezoresistive sensing unit; 4 - electrode lead-out end; 5 - flexible silicone substrate.

[0030] Figure 2 is a flowchart of the active protection method for robots based on three levels of pain sensitivity described in this invention;

[0031] Figure 3 is a schematic diagram of the density layout of the whole-body bionic skin of the robot described in this invention;

[0032] Figure 4 is a schematic diagram of the foot sensor layout and gyroscope-assisted walking balance adjustment logic described in this invention. Detailed Implementation

[0033] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides detailed implementation methods and specific operating steps; however, the scope of protection of the present invention is not limited to the following embodiments.

[0034] The present invention discloses a biomimetic flexible tactile sensor based on pressure pain sensation and an active protection method for robots, which can be widely used in humanoid robots, industrial robotic arms, rehabilitation prostheses, special operation robots and other fields.

[0035] Example 1: Fabrication and Structure of a Bionic Flexible Tactile Sensor

[0036] The biomimetic flexible tactile sensor described in this embodiment is fabricated using a DIW direct-write molding process, as illustrated in Figure 1 below:

[0037] 1. Material Formulation: Three types of silicone rubber-based inks of the same system are formulated to ensure complete matching of thermal expansion coefficients and avoid interface debonding.

[0038] Flexible silicone-based ink: an addition-cured two-component silicone rubber whose hardness can be adjusted according to the application area;

[0039] Flexible conductive silicone ink: Silicone rubber matrix doped with carbon nanotubes and graphene composite conductive filler, conductivity ≥100S / m;

[0040] Piezoresistive conductive silicone ink: The silicone rubber matrix is ​​doped with high specific surface area conductive carbon black and carbon nanotube composite filler, which has linear piezoresistive characteristics.

[0041] 2. Integrated printing process:

[0042] Substrate printing: Using DIW 3D printing equipment, a flexible silicone substrate is printed on a pre-set curved mold. The substrate thickness is 0.5~2mm, which perfectly fits the curved shape of the corresponding part of the robot. After printing, it is pre-cured for 5~10 minutes.

[0043] Conductive mesh printing: Print a cross-shaped conductive mesh on a pre-cured substrate. The row lines intersect the column lines perpendicularly. The line width is 0.1~0.5mm, and the line spacing is set to 0.3~10mm according to the density requirements. After printing, a second curing is performed.

[0044] Piezoresistive sensing unit printing: At each intersection of the row and column conductive grid, a spherical cap-shaped micro-convex piezoresistive sensing unit is printed with a bottom diameter of 0.2~0.8mm, a convex height of 0.05~0.2mm, and a height-to-diameter ratio of 1:4~1:8. After printing, the entire unit is cured to complete the sensor fabrication.

[0045] Electrode leads: Electrodes are printed at the ends of rows and columns of the conductive grid and connected to the back-end acquisition circuit via flexible ribbon cables.

[0046] 3. Structure and working principle:

[0047] Flexible silicone substrate 5: This serves as the core carrier of the sensor, providing physical support, electrical insulation, and environmental protection for the embedded conductive mesh;

[0048] Row conductive line 1 and column conductive line 2: They are embedded in the substrate perpendicularly to each other, forming an M×N matrix. The row lines serve as scanning gating lines, and the column lines serve as signal reading lines.

[0049] Micro-convex piezoresistive sensing unit 3: spherical crown structure, the bottom completely covers the row and column intersection points, and is stably connected with the row and column conductive lines; when not under pressure, it is in a high resistance state, and there is no effective conductive path between the row and column lines; when under pressure, it is uniformly compressed, and the internal conductive fillers come into contact with each other to form a conductive path, and the resistance decreases linearly with the pressure, realizing force-to-electricity conversion.

[0050] 4. Matching acquisition circuit: The row and column terminals of the conductive grid are connected to multiple analog switches and decoders, sharing a single 24-bit high-precision ADC. Force data acquisition of all sensing units is achieved through time-division scanning, eliminating the need for a large number of chips and complex wiring.

[0051] Example 2: Precise Positioning Method for Pressure Points

[0052] This embodiment is based on the sensor described in Embodiment 1, and achieves accurate positioning of the pressure point through a row and column matrix time-division scanning method. The specific steps are as follows:

[0053] 1. Initialization configuration: The main control MCU initializes all GPIO pins, analog switches and ADC, sets all row conductive lines to the off state, and puts all column conductive lines into signal acquisition standby state;

[0054] 2. Row-by-row selection: The main controller selects the i-th row sequentially according to the row number (i=1,2,3...M, where M is the total number of rows), applies a constant excitation voltage of 3.3V to the selected i-th row, and sets all other row lines to a high-impedance off state to avoid signal crosstalk between row lines;

[0055] 3. Column signal synchronous acquisition: After the i-th row is selected, the main controller sequentially acquires the output voltage value of the j-th column (j=1,2,3...N, where N is the total number of columns) through the ADC, records the voltage data corresponding to each column, and after completing the full column scan of the i-th row, the i-th row is turned off and the next row is selected, until the full matrix scan of all rows is completed;

[0056] 4. Resistance to Pressure Conversion: For each coordinate (i,j), based on the collected column voltage value, the real-time resistance value of the piezoresistive sensing unit at that coordinate is calculated using Ohm's law.

[0057]

[0058] in, The value of the sampling resistor at the column end. This is the horizontal drive voltage. The output voltage collected in column j is used as an example; then, the real-time resistance value is converted into the real-time force F of the coordinate sensing unit through the pre-calibrated "resistance-pressure correspondence curve".

[0059] 5. Pressure point determination and coordinate locking: The pressure trigger resistor threshold is pre-calibrated to 80% of the initial resistance value of the sensing unit when it is not under pressure. When the real-time resistance value of a certain coordinate is lower than the threshold, the coordinate is determined to be a pressure point, and row number i and column number j are locked to complete the precise positioning of the pressure point.

[0060] This embodiment employs the reverse bias method of the row line and the characteristics of the high-resistivity unpressurized unit to completely solve the "ghost point" crosstalk problem of the resistance matrix scanning, and can achieve synchronous positioning of single-point / multi-point pressure with a positioning accuracy of ±0.05mm, providing core support for pain trigger position locking.

[0061] Example 3: Active Protection Method for Robots Based on Three-Level Pain Sensation Values

[0062] This embodiment, based on the sensor described in Embodiment 1 and the positioning method described in Embodiment 2, achieves graded active protection for the robot, as illustrated in Figure 2 below:

[0063] 1. Parameter pre-calibration:

[0064] The sensor and the corresponding robot body structure were pressure tested by a universal pressure testing machine to measure the critical pressure BP at which the structure undergoes irreversible deformation and functional failure.

[0065] Calibrate a three-level progressive threshold: pre-trigger threshold = 0.5 × BP, trigger threshold = 0.8 × BP, damage value = 1.0 × BP;

[0066] Pre-calibrate the "resistance-pressure correspondence curve" and the pressure-triggered resistance threshold.

[0067] 2. Complete execution process:

[0068] Step S1: The robot's main controller acquires the pressure data of each sensing unit in real time through the acquisition circuit, calculates the real-time pressure at a single point P=F / S, and substitutes it into the formula to calculate the robot's pain value Pain=(P / BP)×100.

[0069] Step S2: Compare the real-time pain level with the three-level threshold to determine the risk level.

[0070] When 0 < Pain < 50, it is determined to be risk-free, and only normal tactile information is output, which is collected in a loop.

[0071] When 50≤Pain<80, it is determined that the pre-trigger threshold has been reached, and there is a potential risk of damage.

[0072] When 80≤Pain<100, it is determined that the trigger threshold has been reached and the structure is on the verge of damage;

[0073] When Pain ≥ 100, it is determined that the damage value has been reached and the structure has suffered irreversible damage.

[0074] Step S3: Implement graded protection:

[0075] Pre-triggered threshold corresponds to early warning and protection actions: control the robot to decelerate, reduce the applied pressure, send an early warning signal to the host computer, and avoid high-risk contact in advance;

[0076] The trigger threshold corresponds to the core active protection action: the pain trigger location is locked by the method in Example 2. If it is triggered by the hand, the robotic arm is controlled to grasp and wear a flexible protective glove, spreading the single point of force on the fingertip to the entire fingertip area; if it is triggered by the foot, protective shoe covers are controlled to be worn to increase the contact area of ​​the foot; if it is triggered by the torso, the posture is adjusted to lie flat to increase the overall contact area of ​​the torso and reduce local pressure from the source.

[0077] Damage value corresponds to emergency protection action: control the robot to immediately stop the current operation, urgently retract the corresponding limb, cut off the power of the actuator, and avoid secondary damage.

[0078] Step S4 Closed-loop recovery: After the protective action is performed, the local pressure and pain value are monitored in real time. When the pain value returns to zero and the pressure drops back to a safe range, the risk is determined to be eliminated, the robot resumes normal operation, and the complete active protection closed loop is completed.

[0079] Example 4: Robotic Full-Body Bionic Skin System

[0080] This embodiment extends the sensor described in Embodiment 1 to the entire robot, forming a full-area bionic skin system, as illustrated in Figure 3 below:

[0081] 1. High-density sensing area ①: Distributed in the robot's fingertips, palms, forefoot and heel, with a sensing unit spacing of 0.5mm and a point density of 400 points / cm², used for fine operation, texture recognition and center of gravity detection; threshold coefficients are set to α1=0.4, α2=0.7 and α3=1.0, with the highest pain sensitivity, triggering protection in advance.

[0082] 2. Medium-density sensing area ②: Distributed on the back of the robot's fingers, forearm, neck, and lower leg, with a sensing unit spacing of 2mm and a point density of 25 points / cm², used for routine contact and collision sensing; threshold coefficients are set to α1=0.5, α2=0.8, and α3=1.0 to balance sensitivity and operational continuity.

[0083] 3. Low-density sensing area ③: Distributed on the robot's upper arm, torso, back, and thighs, with a sensing unit spacing of 5mm and a point density of 4 points / cm², used for large-area collision protection; threshold coefficients are set to α1=0.6, α2=0.9, and α3=1.1 to avoid frequent triggering of protection due to slight contact.

[0084] All sensors on the robot's body share a single acquisition circuit and pain calculation algorithm. Time-division scanning is achieved through multiple analog switches. When pain is triggered in any area, corresponding autonomous protective actions can be executed, forming a complete whole-body bionic perception and protection system.

[0085] Example 5: A method for adjusting walking balance using foot sensors and gyroscopes

[0086] This embodiment focuses on the robot's foot application scenario, employing the sensor described in Embodiment 1 and fusing data from the robot's gyroscope to achieve walking balance adjustment. The following explanation is based on Figure 4:

[0087] 1. Initial balancing reference calibration

[0088] After the robot is powered on, a three-level static calibration process is performed:

[0089] Bipedal Standing Calibration: Control the robot to maintain a stationary, upright, bipedal supporting posture, collect the plantar pressure distribution, and calculate the coordinates of the stable center of gravity of both feet. Set a threshold for preventing both feet from tilting. Simultaneously acquire the reference pitch angle θ0, roll angle ϕ0, and yaw angle ψ0 from the six-axis gyroscope on the fuselage, and set the bipedal attitude safety thresholds: pitch angle ±10° and roll angle ±5°.

[0090] Single-leg standing calibration: Control the robot to lift its left and right feet respectively to complete the calibration of single-leg support on the corresponding side, collect the plantar pressure distribution, and calculate the coordinates of the single-leg stable center of gravity. Set a threshold for preventing tilting on one foot; synchronously collect the reference attitude angle of one foot from the gyroscope, and set the safety threshold for one foot attitude: pitch angle ±8°, roll angle ±3°.

[0091] 2. Dynamic balance adjustment of walking gait (taking the stepping out with the right foot as an example)

[0092] Stepping preparation phase: With both feet supporting the weight, slowly shift the center of gravity to the left supporting foot while simultaneously monitoring posture changes via the gyroscope; if the gyroscope detects that the pitch / roll angle exceeds the posture threshold of both feet, immediately pause the weight shift and adjust the posture to pull back; until the pressure ratio of the left foot increases to 70%~80%, preparing for the right foot to lift.

[0093] Left single-leg support phase: The right foot is fully lifted, changing to a left single-leg support state, with the force concentrated on the left forefoot; the main control switches to the left single-leg balance benchmark, using Kalman filtering to fuse foot center of gravity data and gyroscope attitude data to comprehensively judge stability; if the center of gravity shifts or the attitude angle exceeds the threshold, the angles of the left hip, left knee, and left ankle joints are adjusted to pull the center of gravity back to a safe range; if a sudden increase in local pressure on the left forefoot triggers pain, the landing rhythm of the right foot is accelerated while ensuring posture stability.

[0094] Right foot swing phase: The right foot swings forward, and the main controller detects the attitude angular velocity through the gyroscope, predicts the attitude change trend, and adjusts the left foot attitude in advance to suppress instability.

[0095] Right foot landing phase: The front part of the right foot lands first, gradually transitioning to the entire foot, restoring the support state of both feet; the main control switches back to the balance benchmark of both feet, smoothly transitioning the center of gravity and posture to an upright state, completing one stepping cycle.

[0096] 3. Emergency protection: If the center of gravity shifts or the posture angle exceeds the threshold, immediately trigger the emergency anti-tipping action and quickly lower the right foot to restore support with both feet; if a sudden increase in local pressure on the sole of the foot triggers pain, adjust the landing angle or wear protective shoe covers to avoid structural damage.

[0097] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for active protection of a robot, characterized in that, Includes the following steps: S1. Using the biomimetic flexible tactile sensor mounted on the robot, the force data of each sensing unit is collected in real time. The real-time pressure of a single point or area is calculated based on the force per unit area, and the robot's pain value is calculated based on the real-time pressure. The robot's pain value is a quantitative technical parameter characterizing the risk of damage to the robot's body structure. S2. Based on the robot's structural characteristics, a three-level progressive threshold is pre-set for the robot's pain sensitivity value, namely, pre-trigger threshold, trigger threshold, and damage value; S3. Based on the real-time calculated pain level of the robot and the matching relationship with the corresponding threshold, determine the structural damage risk level of the corresponding part and execute the corresponding graded protective actions: When the robot's pain sensitivity reaches the pre-triggered threshold, it determines that there is a potential risk of structural damage to the corresponding part and executes a warning and protective action. When the robot's pain sensitivity reaches the trigger threshold, it determines that the corresponding part of the structure is on the verge of damage and executes the core active protection action; When the robot's pain sensitivity reaches the damage threshold, it determines that the corresponding structural part has suffered irreversible damage and executes an emergency shutdown protection action.

2. The active protection method for a robot according to claim 1, characterized in that, The specific method for calculating the robot's pain sensitivity value is as follows: The critical pressure BP for irreversible damage to the biomimetic flexible tactile sensor and the corresponding robot body structure is pre-calibrated, where BP is the minimum pressure value at which the structure undergoes irreversible deformation and functional failure. Real-time calculation of the real-time pressure of a single point or region: P=F / S, where F is the real-time force on the sensing unit / region and S is the effective force-bearing area of ​​the sensing unit / region; The robot's pain value Pain = (P / BP) × 100, and the pain value is positively correlated with real-time pressure and the degree of structural damage risk.

3. The active protection method for a robot according to claim 2, characterized in that, The calibration rule for the three-level progressive threshold is as follows: Pre-trigger threshold = α1 × BP, where the proportional coefficient α1 ranges from 0.4 ≤ α1 ≤ 0.6; Trigger threshold = α2 × BP, where the proportionality coefficient α2 ranges from 0.6 ≤ α2 ≤ 0.9; Damage value = α3 × BP, where the proportionality coefficient α3 ≥ 1.

0.

4. The active protection method for a robot according to claim 1, characterized in that, The specific tiered protection actions are as follows: The early warning and protection actions include: controlling the robot to decelerate, reducing the applied pressure, adjusting the posture to avoid high-risk contact, and sending a structural damage early warning signal to the host computer at the same time. The core active protection actions include: controlling the robot to autonomously perform protective actions that increase the force-bearing area, specifically at least one of the following: wearing flexible protective gloves / shoe covers, pressing multiple fingers together to distribute pressure, wrapping with flexible cushioning material, and adjusting posture to increase the contact area of ​​the torso, thereby reducing local pressure from the physical source; The emergency protection actions include: controlling the robot to immediately stop the current operation, urgently retracting the corresponding limbs, and cutting off the power to the corresponding actuators to avoid secondary structural damage and safety accidents.

5. The active protection method for a robot according to claim 1, which is a parallel dependent claim 1, characterized in that, In step S3, the two-dimensional coordinates of the pain trigger location are located using a row and column matrix time-division scanning method. The specific steps are as follows: Initialize the configuration by setting all row conductive lines of the sensor to the off state and all column conductive lines to the signal acquisition standby state. Select the i-th row in order of row number, apply a constant excitation voltage to the selected i-th row, and set all other row lines to a high-impedance off state; After selecting row i, the output voltage value of column j is collected synchronously. After completing the full column scan of row i, row i is turned off, and the next row is selected, until the full matrix scan of all rows is completed. Based on the collected column voltage value, the real-time resistance value and force value of the sensing unit at the corresponding coordinate (i,j) are calculated. When the real-time resistance value is lower than the preset pressure trigger resistance threshold, the coordinate (i,j) is determined to be the pain trigger point, and the position is locked.

6. The active protection method for a robot according to claim 1, characterized in that, The biomimetic flexible tactile sensors are distributed across different parts of the robot's body, with different sensor unit density layouts and different three-level thresholds for each part: For the delicate operation and weight-bearing areas of fingertips, palms, and soles, a high-density sensor unit layout is used, with a sensor unit spacing of 0.3mm~1.0mm. The proportional coefficients α1, α2, and α3 are taken as the lower limit of the range, and the threshold is more conservative. The conventional sensing areas, such as the back of the fingers, forearm, neck, and lower leg, adopt a medium-density sensing unit layout with a sensing unit spacing of 1.0mm to 3.0mm, and the proportional coefficients α1, α2, and α3 are taken as the midpoint of the range. The large protective area covering the upper arm, torso, back, and thighs uses a low-density sensor unit layout with a sensor unit spacing of 3.0mm to 10.0mm. The proportional coefficients α1, α2, and α3 are taken as the upper limit of the range to avoid frequent triggering of protection by slight contact.

7. The active protection method for a robot according to claim 1, characterized in that, When the biomimetic flexible tactile sensors are distributed on the soles of the robot's feet, a walking balance adjustment method coordinated with gyroscopes is also included, with the following specific steps: S0. Initial balance reference calibration: Complete the reference calibration for bipedal standing and left and right single-leg standing respectively, obtain the stable center of gravity reference and non-tilting threshold for bipedal / single-leg standing, and simultaneously collect the reference attitude angle and attitude safety threshold of the built-in gyroscope. S1. During walking, simultaneously collect data on the pressure distribution across the entire sole of the foot and real-time attitude angle and angular velocity data from the gyroscope; S2. By fusion of data, calculate the current center of gravity coordinates and offset, real-time attitude angle and deviation, and comprehensively determine whether it is within the safe range; S3. If within a safe range, dynamically adjust the angles of the hip, knee, and ankle joints to correct the walking gait and maintain stable walking; if outside the safe range, perform emergency anti-tipping actions; if a sudden increase in local pressure on the sole of the foot triggers pain, perform emergency foot lifting and adjustment of the landing point while ensuring postural stability.

8. The active protection method for a robot according to claim 7, characterized in that, Taking the robot's walking gait where it first steps with its right foot as an example, the specific steps for balance adjustment are as follows: Stepping preparation phase: With both feet supporting the weight, control the shift of the center of gravity to the left supporting foot, and simultaneously monitor the posture using a gyroscope. If the posture exceeds the safe threshold, pause the shift until the pressure on the left foot reaches 70% to 80%. Left single-leg support phase: After the right foot is lifted, switch to the left single-leg balance benchmark, integrate the foot center of gravity data and gyroscope posture data to comprehensively judge stability, adjust the angles of the left hip, left knee and left ankle joints to keep the center of gravity within a safe range; if a sudden increase in local pressure on the left forefoot triggers pain, speed up the landing rhythm of the right foot while ensuring posture stability. Right foot swing phase: The gyroscope detects the attitude angular velocity, predicts the attitude change trend, and adjusts the left foot attitude in advance to suppress instability; Right foot landing phase: After the right foot lands, restore support with both feet, switch back to the balance benchmark with both feet, and smoothly transition the center of gravity and posture to an upright state, completing one stepping cycle.

9. A biomimetic flexible tactile sensor based on pressure-induced pain sensation, used to implement the robot active protection method according to any one of claims 1-8, characterized in that, The sensor is fabricated using a direct-write molding (DIW) integrated printing process. The overall structure is a fully flexible structure without layers or splicing, including a flexible silicone substrate, a cross-shaped conductive grid with multiple analog switches and decoders connected to the row and column ends, and a micro-convex piezoresistive sensing unit. The flexible silicone substrate is a one-piece molded flexible structure that can adapt to the curved shape of the corresponding part of the robot. The cross-shaped conductive grid is printed from flexible conductive silicone and embedded inside the flexible silicone substrate, and is cured integrally with the substrate. The micro-convex piezoresistive sensing unit has a spherical crown-shaped structure and is located at the intersection of rows and columns of the cross-shaped conductive grid. It is made of piezoresistive conductive silicone material, and its resistance value changes continuously and linearly with the applied pressure.

10. The biomimetic flexible tactile sensor according to claim 9, characterized in that, The bottom diameter of the micro-convex piezoresistive sensing unit is 0.2mm~0.8mm, the convex height is 0.05mm~0.2mm, and the height-to-diameter ratio is 1:4~1:8; Its bottom completely covers the intersection of the row and column conductive grids, forming a stable electrical connection with the row and column conductive lines.

11. The biomimetic flexible tactile sensor according to claim 9, characterized in that, The flexible silicone substrate uses a differentiated material formulation: The sensors used for fingertips and palms are made of soft, highly sensitive silicone with a Shore 00-30 to Shore A10 substrate. The sensor used on the sole of the foot uses a substrate made of high wear-resistant and high impact-resistant silicone composite material with Shore A10 to Shore A30. The sensor used for the torso uses a general-purpose flexible silicone substrate of Shore A10 to Shore A30.

12. The biomimetic flexible tactile sensor according to claim 9, characterized in that, The cross-shaped conductive grid has multiple analog switches and decoders connected to its row and column ends, sharing a single high-precision ADC. It achieves force data acquisition for all sensing units through time-division scanning.

Images