Variable-stiffness flexible mechanical arm based on shape memory alloy and driving method of variable-stiffness flexible mechanical arm

Abstract

The invention discloses a variable-stiffness flexible mechanical arm based on shape memory alloy and a driving method of the variable-stiffness flexible mechanical arm, and relates to the technical field of flexible robots. Integrated closed-loop control over variable stiffness and bending driving of the flexible mechanical arm is achieved, specifically, the main controller module accurately solves the real-time stiffness and deformation quantity of an SMA driving wire through a mapping model according to received target stiffness and target bending angle instructions and in combination with a real-time resistance value returned by the signal acquisition module, and the real-time stiffness and deformation quantity of the SMA driving wire is calculated through the mapping model. According to the design, the control precision is improved, the stability and reliability of the mechanical arm in a complex environment are ensured through a real-time feedback mechanism, and the operation efficiency and the task completion quality are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of flexible robotics, specifically to a variable stiffness flexible robotic arm based on shape memory alloys and its driving method. Background Technology

[0002] In the field of flexible robotics, the demand for flexible robotic arms is increasing with the rapid development of medical, industrial inspection, and rescue fields. Flexible robotic arms, due to their unique compliance and adaptability, can perform precise operations in complex environments, making them a current research hotspot. Among them, flexible robotic arms based on shape memory alloys show great potential in variable stiffness and bending actuation due to their unique shape memory effect and hyperelastic properties.

[0003] Traditional SMA-driven flexible robotic arm technology suffers from several shortcomings: First, low control precision. Due to the lack of an effective real-time feedback mechanism, traditional methods struggle to precisely control the deformation and stiffness of the SMA drive filament, resulting in poor stability and reliability of the robotic arm in complex environments. Second, low integration. Traditional methods often separate the control of variable stiffness and bending drive, making it difficult to achieve synergistic optimization and limiting the overall performance of the robotic arm. Furthermore, traditional methods exhibit poor adaptability to environmental temperature changes. The phase transition characteristics of SMA are significantly affected by temperature, making it difficult for traditional control methods to maintain stable control precision under varying temperature conditions. Finally, insufficient safety and human-machine interaction. Traditional methods lack effective collision detection and safety protection mechanisms, making it difficult to ensure the safety of the robotic arm during operation and the user-friendliness of human-machine interaction.

[0004] To address the problems of low control precision, low integration, poor environmental adaptability, and insufficient safety and human-machine interaction in traditional SMA-driven flexible robotic arm technology, this invention proposes a variable stiffness flexible robotic arm based on shape memory alloy and its driving method, which is of particular importance. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a variable stiffness flexible robotic arm based on shape memory alloys and its driving method. It can realize integrated closed-loop control of variable stiffness and bending drive of the flexible robotic arm by using a built-in pre-calibrated SMA resistance-stiffness-deformation mapping model and a high-precision signal acquisition module, which significantly improves the control accuracy and stability. At the same time, by designing a wide temperature range environment adaptive temperature compensation unit and a human-machine interaction dual safety protection control unit, the environmental adaptability and safety of the robotic arm are effectively improved, expanding its application scenarios and operational capabilities.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: On the one hand, a variable stiffness flexible robotic arm based on shape memory alloy, the system comprising the following components: a flexible substrate assembly, multiple sets of shape memory alloy SMA drive wires, a signal acquisition module, and a main controller module;

[0007] The flexible substrate assembly is a columnar flexible structure. The multiple sets of SMA drive wires are symmetrically embedded inside the flexible substrate assembly along the axial direction. Both ends of each set of SMA drive wires are limited and fixed to the two ends of the axial direction of the flexible substrate assembly.

[0008] The input terminal of the signal acquisition module is electrically connected to both ends of each group of SMA drive wires, and is used to acquire the loop electrical signal of each group of SMA drive wires in real time and calculate the real-time resistance value of the SMA drive wires.

[0009] The main controller module is electrically connected to the signal acquisition module and the drive power supply terminals of each group of SMA drive wires. The main controller module has a pre-calibrated SMA resistance-stiffness-deformation mapping model. Based on the received target stiffness and target bending angle commands, combined with the real-time resistance value returned by the signal acquisition module, the main controller module calculates the real-time stiffness and real-time deformation of the corresponding SMA drive wire through the mapping model, and outputs the closed-loop controlled drive signal to the corresponding SMA drive wire, realizing the integrated closed-loop control of the flexible robotic arm's variable stiffness and bending drive.

[0010] Furthermore, the flexible matrix component is fabricated using a biomimetic hierarchical composite structure. The specific implementation process is as follows: First, a flexible matrix blank with an embedded axially reinforcing fiber layer is prepared. Medical-grade addition-curing silicone is used as the main matrix material. An axially reinforcing fiber layer of woven aramid fibers is integrally formed on the inner wall of the matrix blank using a mold casting method. The weaving density of the fiber layer is set to gradually increase from the proximal end to the distal end along the matrix axis to avoid axial tensile deformation of the matrix during the driving process. Second, axial through holes matching the number of SMA driving filaments are processed at predetermined circumferential positions on the matrix blank using laser drilling. The inner wall of the through hole is coated with a silane coupling agent modified interface bonding layer after plasma treatment, which improves the interface bonding force between the SMA drive wire and the inner wall of the through hole of the substrate, and avoids the problem of interface debonding during long-term reciprocating drive. Finally, rigid end caps with electrodes are integrally formed at both ends of the substrate blank by vacuum casting. The interface between the rigid end cap and the flexible substrate is provided with a barbed interlocking structure to ensure the stability of the two ends of the SMA drive wire. At the same time, a biocompatible wear-resistant protective coating is coated on the outer surface of the substrate to improve the smoothness and service life of the robotic arm during cavity operation.

[0011] Furthermore, the SMA drive wire needs to undergo directional martensitic transformation and surface modification treatment before assembly. The specific implementation process is as follows: First, the SMA drive wire is vacuum annealed to eliminate residual stress generated during the drawing process and ensure the consistency of the phase transformation properties of the wire. Second, the annealed SMA drive wire is subjected to directional pre-strain thermal cycling. The wire is fixed on a special fixture and a preset pre-stretch is applied. Multiple heating-cooling thermal cycles are performed in a constant temperature chamber. During each cycle, the heating rate, holding time, and cooling time are strictly controlled. The rate of change is improved by directional domestication to form a stable preferred orientation of martensitic variants within the filament, significantly reducing the nonlinear hysteresis during the phase transformation process and improving the linearity of the resistance-stiffness-deformation mapping relationship. Finally, the domesticated SMA drive filament is surface modified by micro-arc oxidation to generate a dense insulating oxide ceramic layer on the filament surface, avoiding short-circuit interference between multiple sets of SMA drive filaments. At the same time, the oxide layer can improve the bonding force between the filament and the flexible substrate interface, reduce fatigue wear during long-term reciprocating drive, and extend the service life of the filament.

[0012] Furthermore, the signal acquisition module adopts a high-precision differential anti-interference acquisition architecture. The specific implementation process is as follows: First, each group of SMA driver wires is configured with an independent acquisition channel. Each channel is equipped with an independent high-precision constant current source circuit. The constant current source circuit is built using a low-temperature drift reference voltage source and an operational amplifier, outputting a constant test current to the corresponding SMA driver wire to avoid crosstalk interference between multiple channels. Second, a differential voltage amplifier circuit is set at the input terminal of each acquisition channel. The two input terminals of the differential circuit are directly electrically connected to the two ends of the SMA driver wire. An instrumentation amplifier is used to build the differential amplifier circuit, which has the characteristics of high input impedance and high common-mode rejection ratio, effectively filtering out interference from the driver wire. The common-mode interference generated by the dynamic circuit and electromagnetic interference in the environment are precisely collected to obtain the millivolt-level voltage difference across the SMA driver wire. Finally, an active low-pass filter circuit is set at the output of the amplifier circuit to filter out high-frequency noise interference before transmitting it to the high-resolution AD conversion unit. The AD conversion unit adopts a synchronous sampling mode to synchronously collect and convert the voltage signals of all channels, ensuring the synchronization of the acquisition of multiple sets of SMA driver wire resistance signals and avoiding closed-loop control errors caused by sampling timing differences. At the same time, the acquisition module has a built-in self-calibration unit, which automatically performs zero-point calibration and gain calibration after each system power-on, eliminating acquisition errors caused by circuit temperature drift and improving the accuracy and long-term stability of resistance signal calculation.

[0013] Furthermore, the resistance-stiffness-deformation mapping model built into the main controller module employs a Prandtl-Ishlinskii hysteresis compensation unit optimized by an improved ant colony algorithm. The weight parameters of the hysteresis operator are optimized using the ant colony algorithm to construct an accurate inverse hysteresis model, as shown in the formula: ,in For the first After the nth iteration The optimal weight values ​​for each Play hysteresis operator are determined by normalizing the measured data from the pre-calibration test to ensure that the initial weights match the actual phase transition characteristics of the SMA drive wire. The pheromone evaporation coefficient ranges from 0.1 to 0.3. The optimal value is determined through preliminary experiments to avoid premature convergence in the algorithm. For the first After the nth iteration The pheromone concentration corresponding to each hysteresis operator is determined by the fitting contribution of the hysteresis operator in the pre-calibration experiment. The higher the contribution of the operator, the higher the initial pheromone concentration. For the first During the nth iteration The pheromone increment of each hysteresis operator is determined inversely by the fitting error of this iteration; the smaller the fitting error, the larger the pheromone increment. For the first In the nth iteration The fitting residuals of a hysteresis operator This represents the maximum fitting residual in this iteration. The residual influence factor, with a value of 2, is used to control the degree of influence of the residual on the weight update. After iterative optimization through this formula, the optimal hysteresis operator weights that adapt to the phase transition characteristics of the target SMA driving wire can be obtained. The constructed hysteresis inverse model can reduce the mapping error caused by SMA phase transition hysteresis by more than 90%, greatly improve the accuracy of the resistance-stiffness-deformation mapping model, and avoid the problems of poor adaptability and large error of the traditional fixed-weight hysteresis model.

[0014] Furthermore, the main controller module incorporates a collaborative driving timing control unit for multiple sets of SMA drive wires. The specific implementation process is as follows: First, based on the target bending angle and target stiffness commands of the flexible robotic arm, the target deformation and target stiffness values ​​corresponding to each set of SMA drive wires are obtained through kinematic calculations. Then, the target resistance threshold corresponding to each set of SMA drive wires is obtained through inverse solution of the resistance-stiffness-deformation mapping model. Second, the driving timing of multiple sets of SMA drive wires is collaboratively planned. Addressing the tension-contraction coupling problem of the opposite SMA drive wires during bending, the driving process is divided into three stages: pre-adjustment phase, main driving phase, and stable phase holding. In the pre-adjustment phase, a small current is applied to all SMA drive wires for preheating, bringing the wire temperature close to the phase change initiation temperature to avoid asynchronous phase changes during subsequent driving. In the main driving phase, the SMA drive wires on the inner side of the target bending direction are preheated with a small current to ensure the wire temperature is close to the phase change initiation temperature, thus avoiding phase change asynchrony during subsequent driving. Drive wire A receives a drive current at a preset current rise rate, while the opposite SMA drive wire receives a small current to maintain its phase transition state, preventing a sudden change in stiffness caused by passive stretching of the opposite wire. During the stable phase holding phase, when the real-time bending angle and stiffness of the flexible robotic arm approach the target value, the adjustment step size of the drive current is gradually reduced, and the drive current of all SMA drive wires is adjusted synchronously to keep the robotic arm stably in the target pose and stiffness state. Finally, during the driving process, the resistance signal of all SMA drive wires is collected in real time through the signal acquisition module to monitor the phase transition progress of each group of wires in real time. When the phase transition progress of a certain group of wires deviates from the planned timing, the drive current of that group of wires is adjusted in real time to ensure that the phase transition process of multiple groups of SMA drive wires is synchronized, avoiding deformation coupling and pose jitter problems, and significantly improving the stability and positioning accuracy of the flexible robotic arm driving process.

[0015] Furthermore, the closed-loop control algorithm built into the main controller module adopts an adaptive weighted dual closed-loop control algorithm. Addressing the coupling problem between the two control variables, bending angle and stiffness, it constructs an adaptive weighted control variable increment fusion formula to achieve coordinated optimization control of the two control variables. The formula is as follows: ,in For the first The PWM drive control increment is output in each control cycle, and the output range of the control increment is limited to -15% to +15% to avoid drive jitter caused by sudden changes in control quantity; The initial value of the gain coefficient for coordinated control is determined by combining the critical proportional gain method with pre-experimentation. The value range is 0.8~2.2, which is used to match the phase change response characteristics of the SMA drive wire and ensure the stability of the control process. For the first The bending angle deviation for each control cycle is obtained from the difference between the target bending angle and the real-time calculated bending angle. For the first The stiffness deviation for each control cycle is obtained from the difference between the target stiffness and the stiffness calculated in real time. For the first The adaptive weighting coefficient for each control cycle ranges from 0 to 1. Its value is determined in real time by the normalized ratio of the angle deviation and stiffness deviation in the current control cycle. When the angle deviation is greater than the stiffness deviation, the weighting coefficient automatically increases to prioritize the control accuracy of the bending angle. When the stiffness deviation is greater than the angle deviation, the weighting coefficient automatically decreases to prioritize the control accuracy of the stiffness. Through this adaptive weighting formula, the coordinated closed-loop control of the two coupled control quantities of bending angle and stiffness can be realized, solving the problem that traditional single closed-loop control cannot take into account the accuracy of the two control quantities, and significantly improving the dynamic control performance and steady-state control accuracy of the flexible robotic arm.

[0016] Furthermore, the main controller module incorporates a human-machine interface dual safety protection control unit. The specific implementation process is as follows: First, a real-time collision force identification model based on the intrinsic resistance signal of the SMA is constructed. A mapping relationship between the resistance change of the SMA drive wire and the external collision force is established through pre-calibration experiments. During the operation of the robotic arm, the resistance signal of the SMA drive wire is collected in real time through the signal acquisition module, and the abrupt change value of the resistance change is calculated. When the abrupt change value of the resistance exceeds the preset normal drive fluctuation range, an external collision can be identified, achieving real-time collision detection without the need for additional force sensors. Second, a dual safety protection mechanism combining passive compliance and active stiffness reduction is constructed. During the normal operation of the robotic arm, the phase transition degree of the SMA drive wire is actively increased to ensure the stiffness and operational accuracy of the robotic arm. Upon detecting an external collision, a safety protection mechanism is immediately triggered. In the first stage, the main controller module quickly cuts off the drive current of the SMA drive wire, causing it to rapidly reverse phase and return to a martensitic low-stiffness state, achieving millisecond-level active stiffness reduction. In the second stage, the passive compliance characteristics of the flexible matrix further mitigate the impact force of the collision, preventing damage to the interacting object. Finally, a redundant safety protection mechanism is set up. When the system experiences power outages, communication interruptions, or other faults, the SMA drive wire automatically loses its drive current and returns to a low-stiffness state, avoiding safety risks under fault conditions. At the same time, the main controller module has a built-in fault self-checking unit that performs a comprehensive self-check of the system's acquisition circuit, drive circuit, and SMA drive wire status after each power-on. If an abnormality is detected, the drive output is immediately locked to ensure the operational safety of the robotic arm throughout its entire life cycle.

[0017] Furthermore, the main controller module incorporates a wide-temperature-range environmental adaptive temperature compensation unit. The specific implementation process is as follows: First, through pre-calibration experiments, a corrected database of the resistance-stiffness-deformation mapping relationship of the SMA drive wire under different ambient temperatures is established. Within a temperature range of -20℃ to 60℃, a temperature test point is set every 5℃. At each temperature test point, a calibration experiment of the full phase change process is performed on the SMA drive wire. Resistance, stiffness, and deformation data at different temperatures are collected, and the mapping model correction coefficient at each temperature point is fitted and stored in the database of the main controller module. Second, during the operation of the robotic arm, the real-time temperature of the wire and the ambient temperature are calculated through the intrinsic resistance signal of the SMA drive wire, without the need for... An additional temperature sensor is installed. Specifically, the basic resistance value of the SMA drive wire in the absence of drive current is collected and combined with the pre-calibrated resistance-temperature characteristic curve to calculate the current ambient temperature. Finally, based on the calculated ambient temperature, the corresponding correction coefficient is retrieved from the correction database to correct the output of the resistance-stiffness-deformation mapping model in real time. When the ambient temperature deviates from the room temperature calibration temperature, the fitting parameters of the mapping model are automatically adjusted to eliminate the influence of ambient temperature changes on the phase transition characteristics of SMA. This ensures that the robotic arm can maintain stable control accuracy in a wide temperature range environment, avoiding the limitation of traditional solutions that can only work stably at room temperature. This expands the application capabilities of the robotic arm in extreme environments such as outdoor and high and low temperature industrial scenarios.

[0018] On the other hand, a driving method for a variable stiffness flexible robotic arm based on shape memory alloys is characterized by the following specific steps:

[0019] S1. Pre-calibration and mapping model construction: Multi-condition calibration test of the SMA drive wire during the martensitic-austenitic full phase transformation process was carried out. In the wide temperature range of -20℃ to 60℃, the resistance value, axial deformation and output stiffness of the SMA drive wire under the full driving current gradient were collected as a one-to-one correspondence dataset. The Play operator weight of the Prandtl-Ishlinskii hysteresis model was iteratively optimized by the improved ant colony algorithm. The resistance-stiffness-deformation mapping model with adaptive hysteresis compensation was constructed. The optimized mapping model and the temperature compensation coefficient library were synchronously stored in the main controller module.

[0020] S2. Control command reception and calculation: The main controller module receives the target bending angle and target stiffness control commands issued by the user for the flexible robotic arm. It calculates the target axial deformation and target output stiffness threshold corresponding to each SMA drive wire in the circumferential direction through the spatial constant curvature kinematic model. Then, it completes the inverse calculation through the mapping model pre-built in step S1 to obtain the target resistance threshold corresponding to each group of SMA drive wires. At the same time, it completes the coordinated drive timing planning of three stages: pre-adjustment phase, main drive phase, and stable phase holding, based on the tension-contraction coupling characteristics of multiple groups of SMA drive wires.

[0021] S3. Synchronous acquisition of intrinsic self-sensing signals: The signal acquisition module outputs a constant low temperature drift test current to the independent acquisition channel of each group of SMA drive wires. At a synchronous sampling frequency of not less than 100Hz, it acquires the differential voltage signal at both ends of each group of SMA drive wires. After active low-pass filtering and high-resolution analog-to-digital conversion, the real-time resistance value of each group of SMA drive wires is obtained by solving Ohm's law. The synchronous acquisition data of all channels is transmitted back to the main controller module in real time.

[0022] S4. Real-time state calculation and control deviation calculation: The main controller module inputs the real-time resistance value collected in step S3 into the resistance-stiffness-deformation mapping model pre-built in step S1, calculates the real-time axial deformation and real-time output stiffness of each SMA drive wire, and then obtains the real-time spatial bending angle and overall real-time stiffness of the flexible robotic arm through forward kinematics calculation. By comparing with the target control command, the bending angle deviation and stiffness deviation are calculated respectively.

[0023] S5. Adaptive Dual-Closed-Loop Cooperative Drive Control: Based on the bending angle deviation and stiffness deviation obtained in step S4, the main controller module calculates the PWM drive control quantity corresponding to each SMA drive wire through the adaptive weighted dual-closed-loop control algorithm. According to the cooperative drive timing planned in step S2, the module outputs the regulated drive current to the corresponding SMA drive wire through multiple independent H-bridge drive circuits, controls the SMA drive wire to undergo martensitic-austenitic phase transformation, and simultaneously realizes the bending angle adjustment and stepless control of the overall stiffness of the flexible robotic arm.

[0024] S6. Stable Phase Holding and Full-Process Safety Monitoring: The main controller module continuously iterates and optimizes the closed-loop control process from steps S3 to S5 until the real-time bending angle and real-time stiffness of the flexible robotic arm are within the preset allowable error range of the target command, and enters the low-power stable phase holding state. During the full drive process, the main controller module uses the real-time collected SMA resistor signal to identify external collision anomalies and perform system fault self-checks. When an abnormal state is triggered, it immediately executes millisecond-level safety stiffness reduction protection operations to ensure the safety of human-machine interaction and system operation.

[0025] Compared with existing technologies, this variable stiffness flexible robotic arm based on shape memory alloy and its driving method have the following advantages:

[0026] I. This invention achieves integrated closed-loop control of variable stiffness and bending drive of a flexible robotic arm by using a built-in pre-calibrated shape memory alloy resistance-stiffness-deformation mapping model combined with a high-precision signal acquisition module. Specifically, the main controller module accurately calculates the real-time stiffness and deformation of the SMA drive wire through the mapping model based on the received target stiffness and target bending angle commands and the real-time resistance value returned by the signal acquisition module, and outputs the closed-loop control drive signal. This design not only improves control accuracy, but also ensures the stability and reliability of the robotic arm in complex environments through a real-time feedback mechanism, significantly improving work efficiency and task completion quality.

[0027] Second, this invention features a specially designed wide-temperature-range environmental adaptive temperature compensation unit. Through pre-calibration experiments, a correction database of the resistance-stiffness-deformation mapping relationship of the SMA drive wire under different ambient temperatures was established. During the operation of the robotic arm, the system can calculate the ambient temperature in real time and call the corresponding correction coefficient to correct the output results of the mapping model, ensuring that the robotic arm can maintain stable control accuracy in a wide temperature range of -20℃ to 60℃. In addition, the main controller module also has a built-in human-machine interaction dual safety protection control unit. By constructing a collision force real-time identification model based on the intrinsic resistance signal of SMA and a dual safety protection mechanism combining passive compliance and active stiffness reduction, the safety of the robotic arm during operation and the user-friendliness of human-machine interaction are effectively guaranteed. These designs significantly expand the application scenarios of the robotic arm and improve its working ability and safety in extreme environments.

[0028] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0030] Figure 1 This is an overall flowchart of a variable stiffness flexible robotic arm based on shape memory alloy.

[0031] Figure 2 A flowchart of the three-stage collaborative driving timing control of multiple SMA drive wires for a variable stiffness flexible robotic arm based on shape memory alloy;

[0032] Figure 3 This is a flowchart of a flexible robotic arm with dual safety protection for human-machine interaction, based on a shape memory alloy and variable stiffness.

[0033] Figure 4 This is a flowchart of a driving method for a variable stiffness flexible robotic arm based on shape memory alloy. Detailed Implementation

[0034] To further illustrate the technical means and effects of the present invention in achieving its intended purpose, the following detailed description of the specific implementation methods, structures, features, and effects of the present invention, in conjunction with the accompanying drawings and preferred embodiments, is provided below.

[0035] Example

[0036] This embodiment addresses the clinical need for minimally invasive abdominal tumor biopsy surgery. It utilizes the variable stiffness flexible robotic arm based on shape memory alloy described in this invention to perform precise bending positioning, soft tissue avoidance, and tumor biopsy within the confined space of the abdominal cavity. The entire process achieves integrated closed-loop control of variable stiffness and bending actuation. The specific implementation process is as follows:

[0037] The flexible substrate component of the flexible robotic arm used in this embodiment is fabricated using a biomimetic hierarchical composite structure. Medical-grade addition-cure silicone is used as the main substrate material. First, a flexible substrate blank with an embedded axially reinforcing fiber layer is prepared by mold casting. An axially reinforcing fiber layer of woven aramid fibers is integrally formed on the inner wall of the substrate blank. The weaving density of the fiber layer increases gradually from the proximal end to the distal end along the axial direction of the substrate. Then, axial through-holes matching the number of SMA drive fibers are machined at predetermined positions on the circumference of the substrate blank using laser drilling. The inner wall of the through-holes is coated with a silane coupling agent-modified interface bonding layer after plasma treatment. Finally, rigid end caps with electrodes are integrally formed at both ends of the substrate blank by vacuum casting. The interface between the rigid end caps and the flexible substrate has a barbed interlocking structure. Simultaneously, a biocompatible coating is applied to the outer surface of the substrate. The SMA drive wires used are oriented martensitic variant domestication and surface modification treatments before assembly. First, the SMA drive wires are vacuum annealed to eliminate residual stress during the wire drawing process. Then, the annealed SMA drive wires are oriented pre-strain thermal cycling domestication. The wires are fixed on a special fixture and a preset pre-stretch is applied. Multiple heating-cooling thermal cycles are performed in a constant temperature chamber. During each cycle, the heating rate, holding time and cooling rate are strictly controlled. Finally, a dense insulating oxide ceramic layer is generated on the surface of the domesticated SMA drive wires through micro-arc oxidation. Multiple sets of SMA drive wires are symmetrically embedded in the axial through holes of the substrate along the axial direction of the flexible substrate component. The two ends of each set of SMA drive wires are limited and fixed by the rigid end caps at both ends of the axial direction of the flexible substrate component.

[0038] After the robotic arm body assembly was completed, multi-condition calibration tests were conducted on the SMA drive wire to perform a martensitic-austenitic full-phase transformation process. Within a wide temperature range of -20℃ to 60℃, a one-to-one correspondence dataset of the SMA drive wire's resistance, axial deformation, and output stiffness under the full drive current gradient was collected. Using a Prandtl-Ishlinskii hysteresis compensation unit optimized with an improved ant colony algorithm, a resistance-stiffness-deformation mapping model with adaptive hysteresis compensation was constructed. The formula is as follows: ,in For the first After the nth iteration The optimal weight values ​​for the Play hysteresis operator; The pheromone evaporation coefficient; For the first After the nth iteration The pheromone concentration corresponding to each hysteresis operator; For the first During the nth iteration The pheromone increment of a hysteresis operator; For the first In the nth iteration The fitting residuals of a hysteresis operator This represents the maximum fitting residual in this iteration. As a residual influence factor, the optimized mapping model and temperature compensation coefficient library are synchronously stored in the main controller module. After the system is powered on, the self-calibration unit built into the signal acquisition module automatically completes zero-point calibration and gain calibration to eliminate acquisition errors caused by circuit temperature drift. The fault self-checking unit of the main controller module performs a comprehensive self-check on the status of the system's acquisition circuit, drive circuit, and SMA drive wire. If an abnormality is found, the drive output is immediately locked. At the same time, the operating environment temperature is calculated through the intrinsic resistance signal of the SMA drive wire, and the corresponding correction coefficient is called from the correction database to correct the output results of the resistance-stiffness-deformation mapping model in real time.

[0039] During the surgical procedure, the main controller module receives control commands from the surgical host computer system regarding the target bending angle and stiffness of the flexible robotic arm. It calculates the target axial deformation and target output stiffness threshold for each circumferential SMA drive wire using a spatial constant curvature kinematic model. Then, it performs inverse calculations using a pre-constructed resistance-stiffness-deformation mapping model to obtain the target resistance threshold for each group of SMA drive wires. Simultaneously, considering the tension-contraction coupling characteristics of multiple SMA drive wires, it completes the coordinated drive timing planning for three stages: pre-adjustment phase, main drive phase, and stable phase maintenance. The signal acquisition module outputs a constant low-temperature drift test current to the independent acquisition channel of each SMA drive wire. At a synchronous sampling frequency of no less than 100Hz, it acquires the differential voltage signal across each SMA drive wire. After active low-pass filtering and high-resolution analog-to-digital conversion, the real-time resistance value of each SMA drive wire is calculated using Ohm's law. The synchronous acquisition data from all channels is then transmitted back to the main controller module in real time.

[0040] The main controller module inputs the acquired real-time resistance values ​​into a pre-constructed resistance-stiffness-deformation mapping model to calculate the real-time axial deformation and real-time output stiffness of each SMA drive wire. Then, it uses forward kinematics to calculate the real-time spatial bending angle and overall real-time stiffness of the flexible robotic arm. Comparing this with the target control command, it calculates the bending angle deviation and stiffness deviation. Based on these deviations, the main controller module uses an adaptive weighted dual-closed-loop control algorithm to calculate the corresponding PWM drive control quantity for each SMA drive wire. The formula is as follows: ,in For the first The PWM drive control increment output in each control cycle; For cooperative control gain coefficient; For the first The bending angle deviation per control cycle; For the first Stiffness deviation per control cycle; For the first The adaptive weighting coefficients of each control cycle, according to the planned collaborative drive sequence, output the regulated drive current to the corresponding SMA drive wires through multiple independent H-bridge drive circuits, controlling the SMA drive wires to undergo martensitic-austenitic phase transformation, and simultaneously realizing the bending angle adjustment and stepless control of the overall stiffness of the flexible robotic arm. In the pre-adjustment phase stage, a small current is first applied to all SMA drive wires for preheating, bringing the wire temperature close to the phase transformation initiation temperature. In the main drive phase stage, drive current is applied to the SMA drive wires on the inner side of the target bending direction according to a preset current rise rate, while simultaneously... A small current is synchronously applied to the SMA drive wires on the side to maintain their phase transition state. During the stable phase holding stage, when the real-time bending angle and stiffness of the flexible robotic arm approach the target value, the adjustment step size of the drive current is gradually reduced, and the drive current of all SMA drive wires is adjusted synchronously to keep the robotic arm stably in the target pose and stiffness state. During the drive process, the resistance signal of all SMA drive wires is collected in real time through the signal acquisition module to monitor the phase transition progress of each group of wires in real time. When the phase transition progress of a certain group of wires deviates from the planned timing, the drive current of that group of wires is adjusted in real time.

[0041] After the robotic arm reaches the target biopsy location of the intra-abdominal tumor lesion, it continuously iterates and optimizes through a closed-loop control process until the real-time bending angle and real-time stiffness of the flexible robotic arm are within the preset allowable error range of the target command. It then enters a low-power steady-state state and, in conjunction with the end-effector biopsy instrument, completes the tumor tissue sampling operation. During the full-drive and surgical procedures, the main controller module uses real-time acquired SMA resistance signals and a collision force real-time identification model to perform external collision anomaly identification and system fault self-checking. When the resistance mutation value exceeds the preset normal drive fluctuation range, it can identify that the robotic arm has come into contact with normal tissue in the intra-abdominal cavity. Upon impact, the safety protection mechanism is immediately triggered. In the first stage, the main controller module quickly cuts off the drive current of the SMA drive wire, causing the SMA drive wire to rapidly reverse phase transition and return to the martensitic low-stiffness state. In the second stage, the passive compliance characteristics of the flexible matrix itself further mitigate the impact force. When the system experiences power failure, communication interruption, or other faults, the SMA drive wire will automatically lose its drive current and return to the low-stiffness state, ensuring the safety of the surgical operation and the user-friendliness of the human-machine interaction throughout the process. After the surgery is completed, the main controller module gradually reduces the drive current, controlling the robotic arm to return to the initial straight and low-stiffness state, completing the entire surgical procedure.

[0042] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

Claims

1. A variable stiffness flexible robotic arm based on shape memory alloy, characterized in that, It includes the following components: flexible substrate assembly, multiple sets of shape memory alloy SMA drive wires, signal acquisition module, and main controller module; The flexible substrate assembly is a columnar flexible structure. The multiple sets of SMA drive wires are symmetrically embedded inside the flexible substrate assembly along the axial direction. Both ends of each set of SMA drive wires are fixed and limited to the two ends of the axial direction of the flexible substrate assembly. The input terminal of the signal acquisition module is electrically connected to both ends of each group of SMA drive wires, and is used to acquire the loop electrical signal of each group of SMA drive wires in real time and calculate the real-time resistance value of the SMA drive wires. The main controller module is electrically connected to the signal acquisition module and the drive power supply terminal of each group of SMA drive wires. The main controller module has a pre-calibrated SMA resistance-stiffness-deformation mapping model built in. The main controller module calculates the real-time stiffness and real-time deformation of the SMA drive wire through the mapping model based on the received target stiffness and target bending angle commands and the real-time resistance value returned by the signal acquisition module. It then outputs a closed-loop control drive signal to the corresponding SMA drive wire to realize the integrated closed-loop control of the flexible robotic arm's variable stiffness and bending drive.

2. The variable stiffness flexible robotic arm based on shape memory alloy according to claim 1, characterized in that, The flexible substrate component is fabricated using a biomimetic hierarchical composite structure. The specific implementation process is as follows: First, a flexible substrate blank with an embedded axial reinforcing fiber layer is prepared. Medical-grade addition-curing silicone is used as the main substrate material. An axial reinforcing fiber layer of woven aramid fibers is integrally formed on the inner wall of the substrate blank by mold casting. The weaving density of the fiber layer is set to gradually increase from the proximal end to the distal end along the axial direction of the substrate. Second, axial through holes matching the number of SMA drive wires are processed at predetermined circumferential positions on the substrate blank by laser drilling. The inner wall of the through holes is coated with a silane coupling agent modified interface bonding layer after plasma treatment to improve the interface bonding force between the SMA drive wires and the inner wall of the substrate through holes. Finally, rigid end caps with electrodes are integrally formed at both ends of the substrate blank by vacuum casting. The interface between the rigid end caps and the flexible substrate is provided with a barbed interlocking structure. At the same time, a biocompatible wear-resistant protective coating is coated on the outer surface of the substrate.

3. The variable stiffness flexible robotic arm based on shape memory alloy according to claim 1, characterized in that, Before assembly, the SMA driving wire needs to undergo directional martensitic variant domestication and surface modification treatment. The specific implementation process is as follows: First, the SMA driving wire is vacuum annealed to eliminate the residual stress generated during the wire drawing process. Second, the annealed SMA driving wire is subjected to directional pre-strain thermal cycling domestication. The wire is fixed on a special fixture and a preset pre-stretch is applied. Multiple heating-cooling thermal cycles are performed in a constant temperature chamber. During each cycle, the heating rate, holding time, and cooling rate are strictly controlled. Through directional domestication, a stable martensitic variant preferred orientation is formed inside the wire, which greatly reduces the nonlinear hysteresis during the phase transformation process of the wire. Finally, the domesticated SMA driving wire is surface modified by micro-arc oxidation to generate a dense insulating oxide ceramic layer on the surface of the wire.

4. The variable stiffness flexible robotic arm based on shape memory alloy according to claim 1, characterized in that, The signal acquisition module adopts a high-precision differential anti-interference acquisition architecture. The specific implementation process is as follows: First, an independent acquisition channel is configured for each group of SMA driver wires. Each channel is equipped with an independent high-precision constant current source circuit. The constant current source circuit is built using a low-temperature drift reference voltage source and an operational amplifier, outputting a constant test current to the corresponding SMA driver wire. Second, a differential voltage amplifier circuit is set at the input end of each acquisition channel. The two input ends of the differential circuit are directly electrically connected to the two ends of the SMA driver wire. An instrumentation amplifier is used to build the differential amplifier circuit. Finally, an active low-pass filter circuit is set at the output end of the amplifier circuit to filter out high-frequency noise interference before transmitting the signal to a high-resolution AD conversion unit. The AD conversion unit adopts a synchronous sampling mode to synchronously acquire and convert the voltage signals of all channels. At the same time, the acquisition module has a built-in self-calibration unit that automatically performs zero-point calibration and gain calibration each time the system is powered on, eliminating acquisition errors caused by circuit temperature drift.

5. A variable stiffness flexible robotic arm based on shape memory alloy according to claim 1, characterized in that, The main controller module's built-in resistance-stiffness-deformation mapping model employs an improved ant colony algorithm-optimized Prandtl-Ishlinskii hysteresis compensation unit. The ant colony algorithm optimizes the weight parameters of the hysteresis operator, constructing an accurate inverse hysteresis model, as shown in the formula: ,in For the first After the nth iteration The optimal weight values ​​for the Play hysteresis operator; The pheromone evaporation coefficient; For the first After the nth iteration The pheromone concentration corresponding to each hysteresis operator; For the first During the nth iteration The pheromone increment of a hysteresis operator; For the first In the nth iteration The fitting residuals of a hysteresis operator This represents the maximum fitting residual in this iteration. This represents the residual impact factor.

6. The variable stiffness flexible robotic arm based on shape memory alloy according to claim 1, characterized in that, The main controller module has a built-in collaborative drive timing control unit for multiple sets of SMA drive wires. The specific implementation process is as follows: First, according to the target bending angle and target stiffness command of the flexible robotic arm, the target deformation and target stiffness value corresponding to each set of SMA drive wires are obtained through kinematic calculation. Then, the target resistance threshold corresponding to each set of SMA drive wires is obtained through the inverse solution of the resistance-stiffness-deformation mapping model. Secondly, the driving timing of multiple SMA driving wires is planned in a coordinated manner. In response to the tension-contraction coupling problem of the opposite SMA driving wire during the bending driving process, the driving process is divided into three stages: pre-adjustment phase, main driving phase, and stable phase holding. In the pre-adjustment phase, a small current is passed through all SMA driving wires to preheat them so that the wire temperature is close to the phase change initiation temperature. During the main driving phase, a driving current is applied to the SMA driving wire on the inside of the target bending direction at a preset current rise rate, while a small current is applied to the SMA driving wire on the opposite side to maintain its phase change state. During the steady-state phase, when the real-time bending angle and stiffness of the flexible robotic arm approach the target value, the adjustment step size of the drive current is gradually reduced, and the drive current of all SMA drive wires is adjusted synchronously to keep the robotic arm stably in the target pose and stiffness state. Finally, during the driving process, the resistance signal of all SMA drive wires is collected in real time through the signal acquisition module to monitor the phase change progress of each group of wires in real time. When the phase change progress of a certain group of wires deviates from the planned timing, the drive current of that group of wires is adjusted in real time.

7. A variable stiffness flexible robotic arm based on shape memory alloy according to claim 1, characterized in that, The main controller module's built-in closed-loop control algorithm employs an adaptive weighted dual-closed-loop control algorithm. Addressing the coupling issue between the bending angle and stiffness control variables, it constructs an adaptive weighted control variable increment fusion formula to achieve coordinated optimization control of the two control variables. The formula is as follows: ,in For the first The PWM drive control increment output in each control cycle; For cooperative control gain coefficient; For the first The bending angle deviation per control cycle; For the first Stiffness deviation per control cycle; For the first Adaptive weighting coefficients for each control cycle.

8. A variable stiffness flexible robotic arm based on shape memory alloy according to claim 1, characterized in that, The main controller module incorporates a human-machine interface dual safety protection control unit. The specific implementation process is as follows: First, a real-time collision force identification model based on the intrinsic resistance signal of the SMA (Structured Motion Wire) is constructed. A mapping relationship between the resistance change of the SMA drive wire and the external collision force is established through pre-calibration experiments. During the operation of the robotic arm, the resistance signal of the SMA drive wire is collected in real time through the signal acquisition module, and the abrupt change value of the resistance change is calculated. When the abrupt change value of the resistance exceeds the preset normal drive fluctuation range, an external collision can be identified. Second, a dual safety protection mechanism combining passive compliance and active stiffness reduction is constructed. During the normal operation of the robotic arm, the phase transition degree of the SMA drive wire is actively increased to ensure... To ensure the rigidity and operational accuracy of the robotic arm, a safety protection mechanism is immediately triggered upon detecting an external collision. In the first stage, the main controller module quickly cuts off the drive current of the SMA drive wire, causing it to rapidly reverse phase and return to a martensitic low-stiffness state. In the second stage, the passive compliance characteristics of the flexible matrix further mitigate the impact force. Finally, a redundant safety protection mechanism is set up. When the system experiences power failure, communication interruption, or other faults, the SMA drive wire automatically loses its drive current and returns to a low-stiffness state. At the same time, the main controller module has a built-in fault self-checking unit that performs a comprehensive self-check of the system's acquisition circuit, drive circuit, and SMA drive wire status after each power-on. If an abnormality is detected, the drive output is immediately locked.

9. A variable stiffness flexible robotic arm based on shape memory alloy according to claim 1, characterized in that, The main controller module has a built-in wide-temperature-range environmental adaptive temperature compensation unit. The specific implementation process is as follows: First, through pre-calibration tests, a correction database of the resistance-stiffness-deformation mapping relationship of the SMA drive wire under different ambient temperatures is established. Within the temperature range of -20℃ to 60℃, a temperature test point is set every 5℃. At each temperature test point, a calibration test of the full phase change process of the SMA drive wire is carried out. Resistance, stiffness, and deformation data at different temperatures are collected, and the mapping model correction coefficients at each temperature point are fitted and stored in the database of the main controller module. Second, during the operation of the robotic arm, the real-time temperature of the wire and the ambient temperature are calculated by solving the intrinsic resistance signal of the SMA drive wire. Finally, based on the calculated ambient temperature, the corresponding correction coefficients are called from the correction database to correct the output results of the resistance-stiffness-deformation mapping model in real time. When the ambient temperature deviates from the room temperature calibration temperature, the fitting parameters of the mapping model are automatically adjusted.

10. A driving method for a variable stiffness flexible robotic arm based on shape memory alloy, applicable to the variable stiffness flexible robotic arm based on shape memory alloy as described in any one of claims 1-9, characterized in that, The specific steps of this method are as follows: S1. Pre-calibration and mapping model construction: Multi-condition calibration test of the SMA drive wire during the martensitic-austenitic full phase transformation process was carried out. In the wide temperature range of -20℃ to 60℃, the resistance value, axial deformation and output stiffness of the SMA drive wire under the full driving current gradient were collected as a one-to-one correspondence dataset. The Play operator weight of the Prandtl-Ishlinskii hysteresis model was iteratively optimized by the improved ant colony algorithm. The resistance-stiffness-deformation mapping model with adaptive hysteresis compensation was constructed. The optimized mapping model and the temperature compensation coefficient library were synchronously stored in the main controller module. S2. Control command reception and calculation: The main controller module receives the target bending angle and target stiffness control commands issued by the user for the flexible robotic arm. It calculates the target axial deformation and target output stiffness threshold corresponding to each SMA drive wire in the circumferential direction through the spatial constant curvature kinematic model. Then, it completes the inverse calculation through the mapping model pre-built in step S1 to obtain the target resistance threshold corresponding to each group of SMA drive wires. At the same time, it completes the coordinated drive timing planning of three stages: pre-adjustment phase, main drive phase, and stable phase holding, based on the tension-contraction coupling characteristics of multiple groups of SMA drive wires. S3. Synchronous acquisition of intrinsic self-sensing signals: The signal acquisition module outputs a constant low temperature drift test current to the independent acquisition channel of each group of SMA drive wires. At a synchronous sampling frequency of not less than 100Hz, it acquires the differential voltage signal at both ends of each group of SMA drive wires. After active low-pass filtering and high-resolution analog-to-digital conversion, the real-time resistance value of each group of SMA drive wires is obtained by solving Ohm's law. The synchronous acquisition data of all channels is transmitted back to the main controller module in real time. S4. Real-time state calculation and control deviation calculation: The main controller module inputs the real-time resistance value collected in step S3 into the resistance-stiffness-deformation mapping model pre-built in step S1, calculates the real-time axial deformation and real-time output stiffness of each SMA drive wire, and then obtains the real-time spatial bending angle and overall real-time stiffness of the flexible robotic arm through forward kinematics calculation. By comparing with the target control command, the bending angle deviation and stiffness deviation are calculated respectively. S5. Adaptive Dual-Closed-Loop Cooperative Drive Control: Based on the bending angle deviation and stiffness deviation obtained in step S4, the main controller module calculates the PWM drive control quantity corresponding to each SMA drive wire through the adaptive weighted dual-closed-loop control algorithm. According to the cooperative drive timing planned in step S2, the module outputs the regulated drive current to the corresponding SMA drive wire through multiple independent H-bridge drive circuits, controls the SMA drive wire to undergo martensitic-austenitic phase transformation, and simultaneously realizes the bending angle adjustment and stepless control of the overall stiffness of the flexible robotic arm. S6. Stable Phase Holding and Full-Process Safety Monitoring: The main controller module continuously iterates and optimizes the closed-loop control process from step S3 to step S5 until the real-time bending angle and real-time stiffness of the flexible robotic arm are within the preset allowable error range of the target command, and enters the low-power stable phase holding state. During the entire drive process, the main controller module uses real-time collected SMA resistor signals to identify external collision anomalies and perform system fault self-checks. When an abnormal state is triggered, it immediately executes millisecond-level safety reduction protection operations to ensure the safety of human-machine interaction and system operation.

Images