A robot finger control method and device based on space reconstruction

Abstract

The embodiment of the application discloses a robot finger control method and device based on space reconstruction, which comprises the following steps: acquiring the current position, speed and vibration frequency of the robot finger when contacting with a target object; performing phase space reconstruction through a time reversal algorithm, backtracking the motion trajectory of the finger in the past N seconds, extracting the topological features of the motion trajectory, generating the motion mode of the finger, and judging whether the current position of the finger is close to a critical point according to the motion mode to generate a first judgment result; when the first judgment result is yes, dynamically adjusting the avoidance weight of the thumb and the four fingers according to the material characteristics and the motion mode to generate a material-adaptive avoidance path; determining the motion trend of the finger according to the motion mode and the target position generated based on the task demand; and controlling the thumb and the four fingers to move to the target position according to the motion trend and the avoidance path. The application improves the accuracy and adaptability of robot finger control.

Description

Technical Field

[0001] This application relates to the field of robot control technology, and in particular to a robot finger control method, device and robot based on spatial reconstruction. Background Technology

[0002] With the rapid development of robotics technology, robots are increasingly widely used in industries such as manufacturing, healthcare, and services. Robot finger control, as one of the core technologies for robot-environment interaction, directly affects the accuracy and safety of task execution. Current technologies for robot finger control typically rely on preset motion trajectories or simple sensor feedback, making it difficult to adapt to complex dynamic environments or the interaction needs of objects with different materials. For example, when grasping flexible objects, the robot may cause deformation due to a lack of adaptive adjustment; when grasping hard objects, uneven force distribution may affect stability. Furthermore, existing methods often lack in-depth analysis of motion patterns when dealing with the complex nonlinear characteristics of finger movements, making it difficult to accurately determine critical states, thus limiting the flexibility and robustness of robot finger control.

[0003] Therefore, there is an urgent need for an adaptive finger control method that can combine motion trajectory analysis and the material properties of the target object to improve the performance of robots in complex tasks. Summary of the Invention

[0004] This application provides a robot finger control method that enables dynamic adjustment of finger movement trajectory and adaptive generation of avoidance path, thereby improving the accuracy and adaptability of robot finger control.

[0005] This application provides the following solution:

[0006] According to a first aspect, a robot finger control method based on spatial reconstruction is provided. The method includes: acquiring the current position, velocity, and vibration frequency of a robot finger when in contact with a target object, wherein the finger includes a thumb and four fingers excluding the thumb; performing phase space reconstruction using a time reversal algorithm to backtrack the finger's motion trajectory over the previous N seconds, where N ranges from 0.1 seconds to 2 seconds, extracting the topological features of the motion trajectory, generating a motion pattern of the finger, and determining whether the current position of the finger is close to a critical point based on the motion pattern, generating a first judgment result; calculating the material properties of the target object based on the vibration frequency, and when the first judgment result is yes, dynamically adjusting the avoidance weights of the thumb and four fingers based on the material properties and the motion pattern to generate a material-adaptive avoidance path; determining the finger's motion trend based on the motion pattern and a target position generated based on task requirements; and controlling the thumb and four fingers to move to the target position based on the motion trend and the avoidance path.

[0007] According to one achievable method in an embodiment of this application, the phase space reconstruction using the time reversal algorithm includes: constructing the phase space of the motion trajectory based on an adaptively selected time delay and embedding dimension using a delayed embedding method; the extraction of topological features of the motion trajectory to generate the motion pattern of the finger includes: extracting the attractor structure of the trajectory as the topological feature to generate the motion pattern of the finger.

[0008] According to one achievable method in an embodiment of this application, extracting the attractor structure of the trajectory as the topological feature to generate the motion pattern of the finger includes: analyzing the trajectory density distribution of the attractor structure in the phase space within the N-second time window, and extracting the topological invariants of the attractor structure; calculating the Hausdorff distance between the attractor structure and the reference periodic trajectory based on the trajectory density distribution and the topological invariants, and classifying the motion trajectory of the finger into a periodic motion pattern or a chaotic motion pattern according to the Hausdorff distance.

[0009] According to one achievable method in an embodiment of this application, the step of determining whether the current position of the finger is close to a critical point based on the motion pattern and generating a first determination result includes: calculating the trajectory curvature of the current position of the finger; when the motion pattern is a periodic motion pattern, using a first curvature threshold to determine whether the current position is close to a critical point; when the motion pattern is a chaotic motion pattern, using a second curvature threshold higher than the first curvature threshold to determine whether the current position is close to a critical point; and generating the first determination result based on the comparison between the trajectory curvature and the corresponding threshold.

[0010] According to one achievable method in an embodiment of this application, the calculation of the material properties of the target object based on the vibration frequency includes: calculating the elastic coefficient of the target object by analyzing the spectral characteristics of the vibration frequency, and classifying the target object as a hard object or a flexible object according to the elastic coefficient.

[0011] According to one achievable method in an embodiment of this application, the step of dynamically adjusting the avoidance weights of the thumb and the four fingers based on the material characteristic parameters and the motion mode includes: when the target object is a flexible object and the motion mode is a chaotic motion mode, the avoidance weight of the thumb is increased first; when the target object is a hard object and the motion mode is a periodic motion mode, the avoidance weight of the four fingers is increased first.

[0012] According to one achievable method in an embodiment of this application, controlling the thumb and four fingers to move to the target position based on the movement trend and the avoidance path includes: when the movement trend is a releasing trend, prioritizing the movement of the four fingers to the target position; when the movement trend is a clenching trend, prioritizing the movement of the thumb to the target position.

[0013] According to a second aspect, a robot finger control device based on spatial reconstruction is provided. The device includes: a finger data acquisition unit configured to acquire the current position, velocity, and vibration frequency of a robot's finger when in contact with a target object, wherein the finger includes a thumb and four fingers excluding the thumb; a first judgment result generation unit configured to perform phase space reconstruction using a time reversal algorithm, backtracking the finger's motion trajectory over the previous N seconds, where N ranges from 0.1 seconds to 2 seconds, extracting the topological features of the motion trajectory, generating the finger's motion pattern, and determining whether the finger's current position is close to a critical point based on the motion pattern, thereby generating a first judgment result; an avoidance path generation unit configured to calculate the material properties of the target object based on the vibration frequency, and when the first judgment result is yes, dynamically adjusting the avoidance weights of the thumb and four fingers based on the material properties and the motion pattern to generate a material-adaptive avoidance path; a motion trend acquisition unit configured to determine the finger's motion trend based on the motion pattern and a target position generated based on task requirements; and a finger movement unit configured to control the thumb and four fingers to move to the target position based on the motion trend and the avoidance path.

[0014] According to a third aspect, a robot is provided, including the spatial reconstruction-based robot finger control device of the second aspect described above.

[0015] According to a fourth aspect, an electronic device is provided, comprising: one or more processors; and a memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the method described in any one of the first aspects.

[0016] According to the specific embodiments provided in this application, the following technical effects are disclosed:

[0017] This application acquires the current position, velocity, and vibration frequency of a finger upon contact with a target object. It then uses a time-reversal algorithm to reconstruct the phase space, traces the motion trajectory, extracts topological features, generates a finger motion pattern, and identifies critical points. Based on the vibration frequency, it calculates the material properties of the target object, dynamically adjusts the avoidance weights of the thumb and four fingers, generates a material-adaptive avoidance path, and controls the finger to move to the target position according to the motion trend. This method significantly improves the accuracy and adaptability of robot finger control, optimizes the grasping force distribution based on the object's material and motion pattern, reduces the risk of deformation in flexible objects, and enhances the stability of grasping hard objects. It is suitable for fine manipulation tasks in complex dynamic environments.

[0018] Of course, any product implementing this application does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A flowchart illustrating a robot finger control method based on spatial reconstruction provided in an embodiment of this application;

[0021] Figure 2 A structural block diagram of a robot finger control device based on spatial reconstruction provided in an embodiment of this application;

[0022] Figure 3 A schematic block diagram of an electronic device provided in an embodiment of this application. Detailed Implementation

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0024] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “a,” “the,” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0025] It should be understood that the term "and / or" used in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship.

[0026] Depending on the context, the word "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0027] Figure 1 A flowchart of a robot finger control method based on spatial reconstruction provided in this application embodiment is shown below. Figure 1 As shown, the method may include the following steps:

[0028] Step 101: Obtain the current position, speed, and vibration frequency of the robot's fingers when in contact with the target object. The fingers include the thumb and the four fingers excluding the thumb.

[0029] Step 102: Reconstruct the phase space using the time reversal algorithm, trace back the movement trajectory of the finger N seconds ago, where N ranges from 0.1 seconds to 2 seconds, extract the topological features of the movement trajectory, generate the movement pattern of the finger, and determine whether the current position of the finger is close to the critical point based on the movement pattern, generating a first judgment result.

[0030] Step 103: Calculate the material characteristic parameters of the target object based on the vibration frequency. When the first judgment result is yes, dynamically adjust the avoidance weights of the thumb and four fingers according to the material characteristic parameters and the motion mode to generate a material-adaptive avoidance path.

[0031] Step 104: Determine the movement trend of the finger based on the movement pattern and the target position generated based on task requirements.

[0032] Step 105: Based on the movement trend and the avoidance path, control the thumb and four fingers to move to the target position.

[0033] As can be seen from the above process, this application obtains the current position, velocity, and vibration frequency of the finger when in contact with the target object, combines this with a time-reversal algorithm to reconstruct the phase space, traces the motion trajectory, extracts topological features, generates the finger motion pattern, and determines the critical point. Based on the vibration frequency, the material properties of the target object are calculated, the avoidance weights of the thumb and four fingers are dynamically adjusted, a material-adaptive avoidance path is generated, and the finger is controlled to move to the target position according to the motion trend. This method significantly improves the accuracy and adaptability of robot finger control, optimizes the grasping force distribution according to the object material and motion pattern, reduces the risk of deformation of flexible objects, and improves the grasping stability of hard objects, making it suitable for fine manipulation tasks in complex dynamic environments.

[0034] The following describes in detail each step of the above process and the effects that can be further produced, with reference to the embodiments. First, the above step 101, namely "obtaining the current position, speed and vibration frequency of the robot's finger when in contact with the target object, wherein the finger includes the thumb and the four fingers other than the thumb", will be described in detail with reference to the embodiments.

[0035] First, obtaining the current position of the fingers refers to accurately measuring the coordinates of the thumb, index, middle, ring, and little fingers in three-dimensional space using devices such as position sensors or encoders. This positional data reflects the static posture of each finger at a given moment, providing a foundation for subsequent motion trajectory analysis. For example, when grasping an object, the current position data helps the robot determine the relative position of the fingers to the target object, ensuring the accuracy of the action.

[0036] Secondly, acquiring finger velocity involves determining the speed of each finger's movement in space using velocity sensors or differential calculations based on position data. Velocity information reflects the dynamic characteristics of the fingers, such as direction and rate of movement, which is crucial for analyzing finger movement trends and predicting potential collisions or critical states. For example, in rapid grasping tasks, velocity data can help robots avoid collisions or grasping failures caused by excessively rapid movements.

[0037] Finally, acquiring the vibration frequency upon contact with the target object refers to detecting the frequency characteristics of the vibration signal generated when a finger contacts the target object using a contact sensor or vibration sensor. The vibration frequency is closely related to the material and physical properties of the target object; for example, hard objects like metal typically produce high-frequency vibrations, while flexible objects like fabric produce low-frequency vibrations. By analyzing the vibration frequency, the robot can infer the material properties of the target object, such as its elastic coefficient, thus providing a basis for subsequent obstacle avoidance path planning and gripping force adjustment. This information acquisition enables the robot to achieve adaptive control on objects of different materials.

[0038] This application addresses the typical five-finger structure design of a robotic hand, explicitly specifying that the fingers include the thumb and the other four fingers, fully considering the unique role of the thumb in grasping and manipulation. The thumb typically possesses greater flexibility and independence, and its motion patterns and control weights may differ from the other four fingers. Therefore, separately collecting and processing data from the thumb and the other four fingers helps achieve more refined collaborative control.

[0039] The following describes in detail step 102, namely, "reconstructing the phase space using a time reversal algorithm, tracing back the movement trajectory of the finger N seconds ago, where N ranges from 0.1 seconds to 2 seconds, extracting the topological features of the movement trajectory, generating the movement pattern of the finger, and determining whether the current position of the finger is close to the critical point based on the movement pattern, and generating a first judgment result", with reference to the embodiments.

[0040] The time-reversal algorithm is used to trace the movement trajectory of a finger over the previous N seconds, where N ranges from 0.1 seconds to 2 seconds. This time window selection fully considers the dynamic characteristics of the robot's finger movements and the requirements of real-time control. The shorter 0.1 seconds is suitable for rapidly changing motion scenarios, such as high-frequency grasping tasks, while the longer 2 seconds is suitable for analyzing more stable motion patterns, such as slow posture adjustments. The core of the time-reversal algorithm lies in reconstructing the finger's continuous trajectory in the time dimension by reversing the finger's movement history. This backtracking is based not only on current position and velocity data but also on historical data to generate a more comprehensive description of the motion trajectory, laying the foundation for subsequent analysis.

[0041] Phase space reconstruction transforms the traced motion trajectory into a dynamic representation in a high-dimensional phase space. Phase space reconstruction is a nonlinear dynamics analysis method that typically employs delayed embedding techniques to map one-dimensional time-series data (such as the position or velocity of a finger) into a high-dimensional space to reveal the underlying laws of motion.

[0042] In this application, phase space reconstruction can ensure that the trajectory accurately reflects the dynamic characteristics of finger movement by adaptively selecting time delay and embedding dimension. For example, the time delay can be determined using the mutual information method, while the embedding dimension can be calculated using the pseudo nearest neighbor method. Time delay is a key parameter in phase space reconstruction, representing the time interval between selected data points in the time series. The mutual information method analyzes the interdependence between data points at different time intervals in the time series to find an optimal time delay value, enabling the reconstructed phase space to retain the dynamic information of the original data to the greatest extent. Specifically, the mutual information method calculates the mutual information value between the time series and its delayed version, finding the delay point where the mutual information first drops to its minimum value. This time point typically indicates the weakest correlation between data points, thus avoiding redundant information and ensuring that the reconstructed phase space clearly reflects the dynamic characteristics of the system. For example, in the trajectory of a robot finger movement, the mutual information method can help determine a suitable time interval to fully reveal the periodic or chaotic characteristics of the trajectory. This method can capture the nonlinear characteristics of finger movement, such as periodicity or chaos, providing a reliable data foundation for subsequent feature extraction.

[0043] Based on phase space reconstruction, topological features of the motion trajectory are extracted to generate finger motion patterns. Topological features mainly refer to the attractor structure of the trajectory in phase space, i.e., the stable or quasi-stable patterns formed by the motion trajectory in high-dimensional space. The attractor structure can reflect the regularity of finger motion; for example, periodic motion manifests as regular closed trajectories, while chaotic motion manifests as complex non-periodic trajectories. By analyzing the trajectory density distribution and topological invariants of the attractor structure, such as connectivity or invariant dimensions, the system can classify the finger motion trajectory into periodic or chaotic motion patterns. This classification not only reveals the intrinsic laws of finger motion but also provides a crucial basis for dynamic control.

[0044] As one feasible approach, this application extracts the attractor structure of the trajectory as the topological feature to generate the motion pattern of the finger, including: analyzing the trajectory density distribution of the attractor structure in the phase space within the N-second time window, and extracting the topological invariants of the attractor structure; calculating the Hausdorff distance between the attractor structure and the reference periodic trajectory based on the trajectory density distribution and the topological invariants, and classifying the motion trajectory of the finger into a periodic motion pattern or a chaotic motion pattern according to the Hausdorff distance.

[0045] Specifically, attractor structures are stable or quasi-stable forms of trajectory evolution in phase space over long periods, reflecting the essential characteristics of a system's dynamic behavior. In the motion trajectory analysis of a robot finger, attractor structures can manifest as regular closed curves (such as periodic motion) or complex irregular forms (such as chaotic motion). Analyzing the trajectory density distribution of attractor structures in phase space within an N-second time window is a crucial step in generating motion patterns. The trajectory density distribution describes the spatial distribution characteristics of finger motion trajectories in phase space, i.e., the density of trajectories in certain regions. For example, trajectories of periodic motion are usually concentrated in a finite closed region, forming a high-density distribution; while trajectories of chaotic motion may spread across a larger area of ​​phase space, exhibiting a lower and non-uniform density distribution. By analyzing the trajectory density distribution, the system can initially determine the regularity of the motion trajectory, providing data support for further extraction of topological features.

[0046] Topological invariants are quantities that describe the geometric properties of attractor structures, such as connectivity, dimension, or fundamental group, which remain invariant under phase space transformations. By calculating topological invariants, the stability and complexity of attractor structures can be quantified. For example, attractors with periodic motion may have simple topological structures, such as a single closed loop, while attractors with chaotic motion may have higher dimensions or complex fractal structures. These invariants provide a mathematical basis for classifying motion patterns, enabling the system to more accurately describe the dynamic characteristics of finger movements.

[0047] Based on trajectory density distribution and topological invariants, the system calculates the Hausdorff distance between the attractor structure and a reference periodic trajectory to classify motion patterns. The Hausdorff distance is a measure of the geometric similarity between two sets of points, reflecting the difference between the attractor structure and a predefined periodic trajectory template. Specifically, the system compares the current attractor structure with a known periodic trajectory template and calculates their maximum deviation distance in phase space, which is the Hausdorff distance. A smaller Hausdorff distance indicates that the trajectory is close to a periodic pattern, exhibiting regular repetitive motion; a larger distance indicates that the trajectory is more in line with a chaotic pattern, exhibiting nonlinear and unpredictable characteristics. A preset threshold can be used to determine the magnitude of the Hausdorff distance. Based on this distance, the system clearly classifies the finger's motion trajectory into either a periodic motion pattern or a chaotic motion pattern.

[0048] Finally, based on the generated motion pattern, the system determines whether the finger's current position is close to a critical point and generates a first judgment result. A critical point refers to a state where the finger's movement may change significantly, such as an impending collision, grasping failure, or a sudden change in motion direction. By calculating geometric features such as trajectory curvature and combining them with the characteristics of the motion pattern (such as periodicity or chaos), the system can determine whether the current position is close to the critical point. For example, a lower curvature threshold may be used in periodic motion patterns, while a higher threshold is used in chaotic motion patterns to adapt to different dynamic characteristics. The first judgment result provides an important basis for subsequent obstacle avoidance path planning and control strategy adjustments, ensuring that the robot can respond to potential risks in a timely manner.

[0049] As one feasible approach, determining whether the current position of the finger is close to a critical point based on the motion pattern and generating a first determination result includes: calculating the trajectory curvature of the current position of the finger; when the motion pattern is a periodic motion pattern, using a first curvature threshold to determine whether the current position is close to a critical point; when the motion pattern is a chaotic motion pattern, using a second curvature threshold higher than the first curvature threshold to determine whether the current position is close to a critical point; and generating the first determination result based on the comparison between the trajectory curvature and the corresponding threshold.

[0050] Different curvature thresholds are used for critical point determination depending on the motion pattern. When the motion pattern is periodic, the system uses the first curvature threshold. Periodic motion patterns typically correspond to regular, repetitive actions, such as regular grasping or releasing operations, and their trajectories in phase space appear as closed or nearly closed curves. Because the trajectories of periodic motion are relatively stable, their critical points are usually related to small changes in curvature; therefore, the first curvature threshold is set to a low value to capture slight trajectory changes. For example, in a periodic grasping task, if the curvature exceeds the first threshold, it may indicate that the finger is approaching the boundary of the target object, requiring an adjustment of the motion direction.

[0051] To set the curvature threshold, one can first analyze the curvature distribution of periodic motion trajectories in phase space using historical data or simulation experiments to determine the typical range of curvature. Secondly, select a statistical characteristic of the curvature distribution, such as the average curvature plus a standard deviation, as the initial threshold benchmark to cover most curvature changes in normal periodic motion. Then, through experimental verification and optimization, adjust the threshold to balance sensitivity and false alarm rate. For example, in a task involving grasping a hard object, if the curvature exceeds a certain value, it may indicate that the finger is approaching the object's edge. The first curvature threshold might be set to the 90th percentile of the curvature distribution in that scenario, ensuring detection of critical states while avoiding excessive false alarms. The method for setting the second curvature threshold is similar to the first, but focuses on the characteristics of chaotic motion. Specific steps include: first, collecting curvature data of trajectories under chaotic motion patterns through simulation or actual operation, analyzing their distribution characteristics; typically, the curvature distribution of chaotic motion has a larger variance and a higher peak value. Next, a high percentile value of the curvature distribution (e.g., the 95th percentile or higher) is selected as the initial threshold benchmark to capture significant trajectory changes in chaotic motion. Furthermore, considering the unpredictability of chaotic motion, a dynamic adjustment mechanism may be introduced, such as dynamically adjusting the second curvature threshold based on changes in real-time trajectory density or topological invariants, to improve the adaptability of the judgment. For example, in a task involving rapid grasping of flexible objects, the second curvature threshold might be set to 1.5 to 2 times the periodic pattern threshold to ensure the detection of drastic trajectory changes.

[0052] Based on a comparison of the trajectory curvature with a corresponding threshold, the system generates a first judgment result. Specifically, if the curvature at the current position exceeds the threshold of the corresponding motion pattern, the system determines that the finger is approaching the critical point and generates a "approaching the critical point" judgment result; otherwise, it is judged as "not approaching the critical point." This judgment result directly affects subsequent control strategies, such as whether it is necessary to adjust the avoidance path or change the gripping force distribution. For example, when approaching the critical point, the system may prioritize increasing the avoidance weight of the thumb or four fingers to avoid collisions or optimize the gripping effect.

[0053] The following describes in detail step 103, namely, "calculating the material properties of the target object based on the vibration frequency, and when the first judgment result is yes, dynamically adjusting the avoidance weights of the thumb and four fingers according to the material properties and the motion mode to generate a material-adaptive avoidance path," with reference to the embodiments.

[0054] When a robot's finger comes into contact with a target object, the vibration signal captured by contact or vibration sensors exhibits specific frequency characteristics closely related to the object's material. For example, hard objects like metal typically produce high-frequency, rapidly decaying vibration signals, while flexible objects like fabric or rubber produce low-frequency, longer-lasting vibration signals. By analyzing the spectral characteristics of the vibration frequencies, material properties of the target object, such as the elastic modulus or damping coefficient, can be calculated. These parameters reflect the object's hardness, flexibility, or deformation capacity, providing crucial information for subsequent obstacle avoidance path planning. For instance, the system might analyze the vibration signal's spectrum using a Fast Fourier Transform to extract the dominant frequency and attenuation rate, thereby inferring whether the target object is hard or flexible.

[0055] When the initial assessment indicates that the finger's current position is approaching a critical point, the system dynamically adjusts the avoidance weights of the thumb and four fingers based on material properties and motion patterns. The confirmation of a critical point means the finger may face risks of collision, grasping failure, or sudden changes in movement direction, thus requiring timely adjustments to the movement strategy. Avoidance weights refer to the priority or constraint strength assigned to the thumb and four fingers when planning finger movement paths, determining which fingers need to be prioritized for positional adjustments to avoid risks. Material properties determine the object's sensitivity to force; for example, flexible objects require more careful force control to avoid deformation, while rigid objects require a more uniform force distribution to ensure grasping stability. Motion patterns reflect the dynamic behavior of the fingers; for example, periodic motion patterns may require stable grasping force adjustments, while chaotic motion patterns may require more flexible avoidance strategies. By integrating these two aspects of information, the system can specifically optimize the avoidance weights.

[0056] The process of dynamically adjusting the avoidance weights of the thumb and four fingers takes into account their different functions in the robot's hand structure. The thumb typically has higher flexibility and independence, often playing a major role in force transmission during grasping tasks, while the four fingers (index, middle, ring, and little fingers) are more responsible for collaborative grasping and stabilizing objects. Therefore, the adjustment of avoidance weights is differentiated according to the specific scenario. For example, when the target object is flexible and its motion pattern is chaotic, the avoidance weight of the thumb may be increased first to reduce direct pressure on the object, thereby reducing the risk of deformation; when the target object is rigid and its motion pattern is periodic, the avoidance weight of the four fingers may be increased first to optimize the distribution of grasping force and ensure uniform force distribution at the contact point. This dynamic adjustment mechanism allows the robot to flexibly adjust the movement priority of the fingers according to the real-time environment and task requirements.

[0057] Based on the adjusted avoidance weights, the system generates a material-adaptive avoidance path. An avoidance path refers to the trajectory of a finger moving from its current position to a target position, avoiding potential risks (such as collisions or excessive deformation). When generating the path, the system comprehensively considers material properties, motion patterns, and avoidance weights to optimize the path's geometry and motion parameters. For example, for flexible objects, the avoidance path might be designed as a smoother curve to reduce contact forces; for rigid objects, the path might prioritize stability through multi-point contact. Path generation is typically achieved through motion planning algorithms, such as gradient descent-based optimization methods or spline curve interpolation, ensuring that the path satisfies both avoidance requirements and task objectives.

[0058] Specifically, first, the system determines the target position of the fingers based on task requirements and defines the initial conditions for path planning by combining the current position, material properties, and motion pattern. The target position is usually provided by the task planning module, such as the coordinates of the contact point when grasping an object. The current position is obtained through a position sensor, material properties (such as elastic coefficient) are obtained through vibration frequency analysis, and the motion pattern (periodic or chaotic) is determined by phase space reconstruction and Hausdorff distance classification. In addition, the avoidance weights of the thumb and four fingers are dynamically adjusted according to material properties and motion pattern. For example, for flexible objects, the avoidance weight of the thumb may be increased to reduce deformation, while for hard objects, the weight of the four fingers may be increased to optimize the grasping force. This information together constitutes the input conditions for path generation.

[0059] Next, an initial avoidance path is generated using spline curve interpolation. A spline curve is a smooth curve that can generate a continuous and smooth trajectory through a series of control points, suitable for describing the motion path of a robot finger. The specific steps include: First, inserting several intermediate control points between the current position and the target position. The positions of these control points can be determined by geometric constraints, such as avoiding potential collision areas or sensitive areas of the target object. For flexible objects, control points may be placed in areas far from the object's surface to reduce contact forces; for rigid objects, control points may be closer to the contact point to ensure grasping stability. Then, a cubic spline interpolation algorithm is used to generate a smooth curve, ensuring that the path is continuous in space and has a differentiable second derivative, thus guaranteeing the smoothness of the motion. For example, when grasping a flexible object, the path may be designed as a wider arc trajectory to reduce direct pressure on the object.

[0060] After generating the initial path, optimization constraints are introduced to adjust the path to meet material adaptation and obstacle avoidance weight requirements. These constraints include: first, obstacle avoidance weight constraints, adjusting the movement priority of each finger in the path based on the weight distribution of the thumb and four fingers; for example, the thumb, with higher weight, may prioritize avoiding obstacles. Second, material property constraints, setting contact force limits for the path based on the elastic coefficient of the target object; for example, the path for flexible objects must ensure the contact force is below a certain threshold to avoid deformation. Third, motion pattern constraints; for periodic motion, the path must maintain regularity to match a stable grasping rhythm; for chaotic motion, the path must have greater flexibility to cope with unpredictable changes. The optimization process typically employs a gradient descent-based algorithm, defining an objective function that comprehensively considers path length, smoothness, obstacle avoidance effect, and contact force limits. The final path is generated by iteratively optimizing and adjusting the control point positions.

[0061] To further improve the adaptability of the path, a real-time feedback mechanism can be introduced. For example, during path execution, sensors can monitor the distance or contact force between the finger and the object in real time. If a potential collision or deformation risk is detected, the control points can be dynamically adjusted and the spline curve recalculated. For instance, when grasping a flexible object, if the vibration sensor detects an excessively high risk of deformation, the curvature of the path can be temporarily increased to move the finger further away from the object surface. This real-time adjustment can be achieved through online optimization algorithms, such as path replanning based on model predictive control.

[0062] Finally, the generated avoidance path is converted into specific motion commands, controlling the actuators of the thumb and four fingers to move to the target position. These motion commands include the joint angles, velocities, and acceleration parameters of each finger, ensuring that the path execution conforms to the planning requirements. For example, in a periodic motion mode, the thumb and four fingers may move in coordination according to a fixed rhythm; while in a chaotic motion mode, the thumb may prioritize adjustments to cope with dynamic changes. After path execution, the deviation between the actual motion trajectory and the planned path is recorded for subsequent optimization and learning.

[0063] The following describes step 104, namely "determining the movement trend of the finger based on the movement pattern and the target position generated based on task requirements," in detail with reference to the embodiments.

[0064] The target position generated based on task requirements refers to the spatial coordinates that the finger is expected to reach, determined according to the specific task objective. Task requirements are typically provided by a higher-level task planning module. For example, in a grasping task, the target position might be a contact point on an object's surface; in a release task, the target position might be a safe point away from the object. These target positions are usually determined through environmental perception systems (such as visual or tactile sensors) or preset task parameters. For instance, when grasping a hard object, the target positions might be multiple evenly distributed points on the object's surface to optimize the grasping force distribution; while when handling flexible objects, the target positions might be set near the object's edge to reduce the risk of deformation. Determining the target position provides clear directional guidance for the movement trend.

[0065] Determining the finger's movement trend is a core step in combining movement patterns and target positions. Specifically, by comparing the relative positions of the current and target positions and considering the characteristics of the movement pattern, the intention of the finger's movement is inferred. For example, in a periodic movement pattern, if the distance between the current and target positions is small and the trajectory curvature is stable, the movement trend might be determined as "stable approach," indicating that the finger will move towards the target position at a steady speed. In a chaotic movement pattern, if the trajectory curvature changes significantly and the target position involves complex objects, the movement trend might be determined as "dynamic adjustment," indicating that the finger needs to frequently change direction to cope with potential obstacles.

[0066] Determining motion trends typically relies on mathematical models and algorithms. For example, vector analysis can be used to calculate the direction vector from the current position to the target position, and combined with topological features of the motion pattern (such as the trajectory density distribution of attractor structures) to predict dynamic changes in motion. A common approach is to use state estimation algorithms (such as Kalman filtering or particle filtering) to fuse position, velocity, and motion pattern information to predict the finger's motion behavior in the next time step. Furthermore, priority constraints can be introduced based on task requirements. For instance, in a grasping task, the motion trend might prioritize the stable approach of the thumb to ensure grasping force, while in a releasing task, it might prioritize the rapid withdrawal of the four fingers to avoid collisions.

[0067] The following describes step 105, namely "controlling the thumb and four fingers to move to the target position according to the movement trend and avoidance path," in detail with reference to the embodiments.

[0068] The process of controlling the thumb and four fingers to move to a target position requires translating the motion trend and avoidance path into specific motion commands. These commands include the joint angles, velocities, and acceleration parameters of each finger, implemented by actuators (such as servo motors or hydraulic systems) in the robot's control system. Specifically, based on the geometry of the avoidance path, the path is decomposed into a series of discrete control points, each corresponding to the desired position of a finger at a specific time step. Then, combining dynamic information about the motion trend, the system assigns appropriate motion parameters to each control point. For example, under a clenched fist tendency, the thumb might approach the target position at a higher speed to secure the gripping force, while under a loosening tendency, the four fingers might retreat at a faster speed to avoid a collision. This control method is implemented through motion planning algorithms, such as inverse kinematics or trajectory tracking control, ensuring that finger movements conform to path constraints and dynamic requirements.

[0069] This technical feature particularly emphasizes the differentiated handling of the thumb and four fingers in control. The thumb typically possesses greater flexibility and independence in the robotic hand structure, undertaking the primary role in transmitting grasping force, while the four fingers (index, middle, ring, and little fingers) are more responsible for assisting in grasping and stabilizing objects. Therefore, the control strategy assigns different priorities based on the movement trend. For example, when the movement trend is a release tendency, the four fingers are prioritized to move quickly along an avoidance path to the target position to avoid collisions with the target object or other obstacles. This priority control can be achieved by assigning higher velocity weights or shorter response times to the four fingers. Conversely, when the movement trend is a fist-clenching tendency, the system prioritizes the thumb to move to the target position to ensure effective transmission of grasping force, for example, by adjusting the thumb's joint angle to more precisely contact key points of the object. This differentiated control strategy fully utilizes the functional characteristics of the thumb and four fingers, optimizing grasping efficiency and safety.

[0070] The method described in this application can be applied to various scenarios, including but not limited to: First, in industrial automation, this method can be used for precision assembly tasks, such as grasping flexible electronic components or rigid mechanical parts. By using material-adaptive avoidance paths and motion trend control, it optimizes the distribution of grasping force and reduces the risk of deformation or damage. Second, in the field of medical robotics, this method can be applied to surgical robots to precisely control the robotic arm to grasp soft tissue or rigid instruments, dynamically adjusting the movement of the thumb and four fingers to avoid tissue damage while ensuring operational stability. Furthermore, in service robot scenarios, such as catering or housekeeping services, this method can be used to grasp objects of different materials (such as flexible food or hard tableware). By analyzing vibration frequencies and motion patterns, it generates adaptive paths to improve grasping efficiency and safety. These scenarios demonstrate the wide applicability of this method in complex environments.

[0071] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0072] According to another embodiment, a robot finger control device based on spatial reconstruction is provided. Figure 2 A schematic block diagram of a spatially reconstructed robotic finger control device according to one embodiment is shown. Figure 2 As shown, the device 200 includes:

[0073] The finger data acquisition unit 201 is configured to acquire the current position, speed and vibration frequency of the robot's fingers when in contact with a target object, the fingers including the thumb and the four fingers excluding the thumb.

[0074] The first judgment result generation unit 202 is configured to reconstruct the phase space through a time reversal algorithm, trace back the movement trajectory of the finger in the previous N seconds, where N ranges from 0.1 seconds to 2 seconds, extract the topological features of the movement trajectory, generate the movement pattern of the finger, and determine whether the current position of the finger is close to the critical point based on the movement pattern, thereby generating the first judgment result.

[0075] The avoidance path generation unit 203 is configured to calculate the material properties of the target object based on the vibration frequency. When the first judgment result is yes, it dynamically adjusts the avoidance weights of the thumb and four fingers according to the material properties and the motion mode to generate a material-adaptive avoidance path.

[0076] The motion trend acquisition unit 204 is configured to determine the motion trend of the finger based on the motion pattern and the target position generated based on task requirements.

[0077] The finger movement unit 205 is configured to control the thumb and the four fingers to move to the target position according to the movement trend and the avoidance path.

[0078] As an implementable approach, the first judgment result generation unit 202 can be configured to: construct the phase space of the motion trajectory based on the adaptively selected time delay and embedding dimension using a delayed embedding method when reconstructing the phase space through the time reversal algorithm; and can be configured to: extract the attractor structure of the trajectory as the topological feature to generate the motion pattern of the finger when extracting the topological features of the motion trajectory to generate the motion pattern of the finger.

[0079] As one feasible approach, the first judgment result generation unit 202, when extracting the attractor structure of the trajectory as the topological feature to generate the motion pattern of the finger, can be configured to: analyze the trajectory density distribution of the attractor structure in the phase space within the N-second time window, and extract the topological invariants of the attractor structure; calculate the Hausdorff distance between the attractor structure and the reference periodic trajectory based on the trajectory density distribution and the topological invariants, and classify the motion trajectory of the finger into a periodic motion pattern or a chaotic motion pattern according to the Hausdorff distance.

[0080] As one feasible approach, the first judgment result generation unit 202, when determining whether the current position of the finger is close to a critical point based on the motion pattern and generating the first judgment result, can be configured to: calculate the trajectory curvature of the current position of the finger; when the motion pattern is a periodic motion pattern, use a first curvature threshold to determine whether the current position is close to a critical point; when the motion pattern is a chaotic motion pattern, use a second curvature threshold higher than the first curvature threshold to determine whether the current position is close to a critical point; and generate the first judgment result based on the comparison between the trajectory curvature and the corresponding threshold.

[0081] As an implementable approach, the avoidance path generation unit 203 can be configured to calculate the elastic coefficient of the target object by analyzing the spectral characteristics of the vibration frequency, and classify the target object as a hard object or a flexible object according to the elastic coefficient.

[0082] As an implementable approach, the avoidance path generation unit 203 can be configured to dynamically adjust the avoidance weights of the thumb and the four fingers based on the material characteristic parameters and the motion mode as follows: when the target object is a flexible object and the motion mode is a chaotic motion mode, the avoidance weight of the thumb is increased first; when the target object is a rigid object and the motion mode is a periodic motion mode, the avoidance weight of the four fingers is increased first.

[0083] As one possible implementation, the finger movement unit 205 can be configured to control the thumb and four fingers to move to the target position according to the movement trend and the avoidance path: when the movement trend is a releasing trend, the four fingers are preferentially controlled to move to the target position; when the movement trend is a clenching trend, the thumb is preferentially controlled to move to the target position.

[0084] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the device embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments. The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0085] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, use and processing of the relevant data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0086] This application also provides a robot configured to include the aforementioned spatial reconstruction-based robot finger control device.

[0087] In addition, embodiments of this application also provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of the method described in any of the foregoing method embodiments.

[0088] And an electronic device comprising: one or more processors; and a memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the method described in any of the foregoing method embodiments.

[0089] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method described in any of the foregoing method embodiments.

[0090] in, Figure 3The architecture of an electronic device is illustrated, which may include a processor 310, a video display adapter 311, a disk drive 312, an input / output interface 313, a network interface 314, and a memory 320. The processor 310, video display adapter 311, disk drive 312, input / output interface 313, network interface 314, and memory 320 can communicate with each other via a communication bus 330.

[0091] The processor 310 can be implemented using a general-purpose CPU, microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits to execute relevant programs in order to implement the technical solution provided in this application.

[0092] The memory 320 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 320 can store the operating system 321 for controlling the operation of the electronic device 300, and the basic input / output system (BIOS) 322 for controlling the low-level operations of the electronic device 300. Additionally, it can store a web browser 323, a data storage management system 324, and a spatially reconfigurable robot finger control device 325, etc. The aforementioned spatially reconfigurable robot finger control device 325 can be the application program that specifically implements the aforementioned steps in this embodiment. In summary, when implementing the technical solution provided in this application through software or firmware, the relevant program code is stored in the memory 320 and executed by the processor 310.

[0093] Input / output interface 313 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components in the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touch screens, microphones, various sensors, etc., and output devices may include displays, speakers, vibrators, indicator lights, etc.

[0094] Network interface 314 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0095] Bus 330 includes a pathway for transmitting information between various components of the device, such as processor 310, video display adapter 311, disk drive 312, input / output interface 313, network interface 314, and memory 320.

[0096] It should be noted that although the above-described device only shows the processor 310, video display adapter 311, disk drive 312, input / output interface 313, network interface 314, memory 320, bus 330, etc., in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the solution of this application, and does not necessarily include all the components shown in the figures.

[0097] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a computer program product. This computer program product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0098] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A robot finger control method based on spatial reconstruction, characterized in that, The method includes: The current position, velocity, and vibration frequency of the robot's fingers when in contact with a target object are obtained; the fingers include the thumb and the four fingers excluding the thumb. Phase space reconstruction is performed using a time reversal algorithm to trace the finger's movement trajectory N seconds prior, where N ranges from 0.1 seconds to 2 seconds. A delayed embedding method is used to construct the phase space of the movement trajectory based on adaptively selected time delay and embedding dimension. The attractor structure of the movement trajectory is extracted as a topological feature to generate the finger's movement pattern. Based on the movement pattern, it is determined whether the finger's current position is close to a critical point, generating a first judgment result. The extraction of the attractor structure as a topological feature to generate the finger's movement pattern includes: analyzing the trajectory density distribution of the attractor structure in the phase space within the N-second time window and extracting the topological invariants of the attractor structure; calculating the Hausdorff distance between the attractor structure and a reference periodic trajectory based on the trajectory density distribution and topological invariants; and classifying the finger's movement trajectory into a periodic movement pattern or a chaotic movement pattern based on the Hausdorff distance. The material properties of the target object are calculated based on the vibration frequency. When the first judgment result is yes, the avoidance weights of the thumb and the four fingers are dynamically adjusted according to the material properties and the motion mode to generate a material-adaptive avoidance path. Based on the described motion pattern and the target position generated based on task requirements, the movement trend of the fingers is determined; Based on the movement trend and the avoidance path, control the thumb and the four fingers to move to the target position.

2. The method according to claim 1, characterized in that, The step of determining whether the current position of the finger is close to a critical point based on the motion pattern and generating a first determination result includes: Calculate the trajectory curvature of the finger at its current position; When the motion pattern is a periodic motion pattern, the first curvature threshold is used to determine whether the current position is close to the critical point; When the motion mode is a chaotic motion mode, a second curvature threshold higher than the first curvature threshold is used to determine whether the current position is close to the critical point; The first judgment result is generated based on the comparison between the trajectory curvature and the corresponding threshold.

3. The method according to claim 1, characterized in that, The calculation of the material properties of the target object based on the vibration frequency includes: By analyzing the spectral characteristics of the vibration frequency, the elastic coefficient of the target object is calculated, and the target object is classified as a rigid object or a flexible object based on the elastic coefficient.

4. The method according to claim 1 or 3, characterized in that, The step of dynamically adjusting the avoidance weights of the thumb and the four fingers based on the material property parameters and the movement pattern includes: When the target object is a flexible object and the motion mode is a chaotic motion mode, the avoidance weight of the thumb is increased first. When the target object is a hard object and the motion pattern is a periodic motion pattern, the avoidance weight of the four fingers is increased preferentially.

5. The method according to claim 1, characterized in that, The step of controlling the thumb and four fingers to move to the target position according to the movement trend and the avoidance path includes: When the movement trend is a releasing trend, the four fingers are preferentially controlled to move to the target position; when the movement trend is a clenching trend, the thumb is preferentially controlled to move to the target position.

6. A robot finger control device based on spatial reconstruction, characterized in that, The device includes: The finger data acquisition unit is configured to acquire the current position, velocity, and vibration frequency of the robot's fingers when in contact with a target object, wherein the fingers include the thumb and the four fingers excluding the thumb; The first judgment result generation unit is configured to reconstruct the phase space using a time reversal algorithm, backtracking the finger's movement trajectory over the previous N seconds, where N ranges from 0.1 seconds to 2 seconds. Using a delayed embedding method, based on adaptively selected time delay and embedding dimension, it constructs the phase space of the movement trajectory, extracts the attractor structure of the trajectory as a topological feature to generate the finger's movement pattern, and determines whether the finger's current position is close to a critical point based on the movement pattern, generating a first judgment result. The extraction of the attractor structure of the movement trajectory as a topological feature to generate the finger's movement pattern includes: analyzing the motion trajectory density distribution of the attractor structure in the phase space within the N-second time window, and extracting the topological invariants of the attractor structure; calculating the Hausdorff distance between the attractor structure and a reference periodic trajectory based on the motion trajectory density distribution and topological invariants; and classifying the finger's movement trajectory into a periodic movement pattern or a chaotic movement pattern based on the Hausdorff distance. The avoidance path generation unit is configured to calculate the material properties of the target object based on the vibration frequency. When the first judgment result is yes, the avoidance weights of the thumb and four fingers are dynamically adjusted according to the material properties and the motion mode to generate a material-adaptive avoidance path. The motion trend acquisition unit is configured to determine the motion trend of the finger based on the motion pattern and the target position generated based on task requirements; The finger movement unit is configured to control the thumb and the four fingers to move to the target position according to the movement trend and the avoidance path.

7. A robot comprising the spatial reconstruction-based robot finger control device of claim 6.

8. An electronic device, characterized in that, include: One or more processors; And a memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the method according to any one of claims 1 to 5.

Images