Full-dimension self-interference detection method for parallel robot platform

Abstract

The application discloses a full-dimension self-interference detection method for a parallel robot platform, and comprises the following steps: acquiring basic data including six-degree-of-freedom pose information, electric cylinder length and motion state information collected by a data collection layer; judging whether a plurality of risk sources of the parallel robot platform exist interference risks by using the basic information in a core detection layer, performing a hierarchical risk control operation on the parallel robot platform in a response execution layer if there is an interference risk, and otherwise, continuously collecting and detecting; wherein the plurality of risk sources of the parallel robot platform include a Hooke's joint detection risk source, a rod length detection risk source, a singularity detection risk source and an OBB collision detection risk source. The application has strong dynamic working condition adaptability and expandability, and effectively avoids response conflicts and decision delays through multi-module collaborative decision-making, thereby providing a systematic solution for safe, efficient and reliable operation of high-end parallel robots.

Description

Technical Field

[0001] This invention relates to the field of robotic arms, and more particularly to a method for full-dimensional self-interference detection of parallel robot platforms. Background Technology

[0002] With the rapid development of intelligent manufacturing equipment, the requirements for mechanical systems have gradually evolved to include compact structure, high positioning accuracy, and fast dynamic response. Traditional serial mechanisms, limited by their large size, limited load-bearing capacity, and error accumulation, are unable to meet the demands of modern industry for high-precision motion control. Parallel six-degree-of-freedom mechanisms (especially the Stewart platform), due to their advantages such as high control accuracy, strong load-bearing capacity, and no error accumulation, have been widely used in high-end fields such as optical telescopes, aerospace simulation, medical rehabilitation equipment, and precision manufacturing.

[0003] In the practical application of parallel robot platforms, safety protection systems are crucial to ensuring reliable equipment operation. However, existing interference detection technologies suffer from the following major technical bottlenecks:

[0004] (1) Lack of internal interference detection technology for Hooke's hinge

[0005] Traditional detection systems primarily focus on external collisions between links, completely ignoring the mechanical interference risks inherent in the Hooke joint's internal structure. As a critical connecting component in parallel platforms, the Hooke joint's inner and outer rings may experience mechanical interference under extreme poses. Experimental studies show that when the hinge deflection angle exceeds 25°, the collision risk of the Hooke joint on a certain Stewart platform reaches as high as 63%. Current technology lacks precise detection methods for the dual-axis angular constraints of the Hooke joint, making real-time monitoring of the hinge's internal safety status impossible.

[0006] (2) Staticization problem of rod length detection strategy

[0007] Current rod length limit detection methods generally employ fixed threshold strategies, failing to adequately consider the dynamic effects during high-speed operation. During platform acceleration or deceleration, inertial forces and dynamic loads exert additional stress on the electric cylinder, which static thresholds cannot adapt to. Statistical data shows that under high-speed conditions, the false alarm rate of traditional static detection methods increases by 37%, severely impacting system reliability and operational efficiency.

[0008] (3) Limitations on the enclosure of the collision detection system

[0009] Commercial collision detection systems typically only support the detection of preset, fixed models, lacking the ability to dynamically integrate user-defined objects. When users need to introduce new fixtures or tools into the workspace, they must rely on the manufacturer to reprogram and configure them, which takes an average of over 48 hours. This closed nature severely limits the system's flexibility and adaptability, failing to meet the demands of modern smart manufacturing for rapid configuration and flexible production.

[0010] (4) Fragmentation of multi-risk source detection system

[0011] Existing technologies distribute functions such as Hooke's hinge detection, pole length monitoring, and unusual configuration early warning across different independent subsystems, lacking a unified coordination mechanism. This fragmented design results in system response delays exceeding 50ms and may lead to conflicting safety decisions when multiple risk sources occur simultaneously, failing to achieve true comprehensive safety protection. Summary of the Invention

[0012] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method for full-dimensional self-interference detection of parallel robot platforms.

[0013] The objective of this invention is achieved through the following technical solution:

[0014] A first aspect of the present invention provides a method for full-dimensional self-interference detection of a parallel robot platform, comprising the following steps:

[0015] The data acquisition layer acquires basic data including six-degree-of-freedom pose information, electric cylinder length, and motion state information. The core detection layer uses the basic data to determine whether there is interference risk from multiple risk sources in the parallel robot platform. If there is interference risk, the response execution layer performs graded risk control operations on the parallel robot platform. Otherwise, the data acquisition and detection are continuously performed in a loop.

[0016] The parallel robot platform has multiple risk sources, including Hooke's hinge detection risk source, rod length detection risk source, singularity detection risk source, and OBB collision detection risk source.

[0017] Furthermore, regarding the detection of risk sources using the Hooke's hinge, the core detection layer utilizes the aforementioned basic data to determine whether there is interference risk among the multiple risk sources of the parallel robot platform. Specifically, this includes:

[0018] S211, Precise calculation of spatial vectors: Based on the current motion pose of the parallel robot platform, the spatial vector of each Hooke hinge is calculated using the inverse kinematics algorithm;

[0019] S212, Precise Calculation of Biaxial Angles: Based on the spatial vector obtained in step S211, calculate the biaxial angles of each Hooke hinge, including the rotation angle about the Y-axis that reflects the degree of inclination of the Hooke hinge in the vertical plane. And the rotation angle about the X-axis, which reflects the degree of rotation of the Hooke hinge in the horizontal plane. ;

[0020] S213, Asymmetric constraint strategy detection: Based on the numbering characteristics of the Hooke's hinge, an asymmetric constraint strategy is used for interference detection based on the constraint range.

[0021] Furthermore, for the risk sources of the Hooke's hinge detection, if there is an interference risk, a graded risk management operation will be performed on the parallel robot platform at the response execution layer, specifically including:

[0022] The dual-axis angles of each Hooke hinge are monitored in real time. When any angle exceeds the preset constraint range, the corresponding safety response mechanism is immediately triggered.

[0023] S311: Level 1 warning: When the dual-axis angle approaches the preset ratio of the constraint range, the system issues a warning signal to alert the user;

[0024] S312: Secondary speed reduction: When the dual-axis angle exceeds the constraint range but does not reach the danger threshold, the movement speed of the parallel robot platform is automatically reduced to a safe level;

[0025] S313: Level 3 Emergency Stop: When the dual-axis angles significantly exceed the constraint range or a rapid change trend is detected, an emergency stop procedure is immediately executed.

[0026] Furthermore, for the risk sources of pole length detection, the core detection layer uses the aforementioned basic data to determine whether there is interference risk among the multiple risk sources of the parallel robot platform, specifically including:

[0027] S221, Two-way Length Precision Verification: To ensure the accuracy of rod length calculation, a two-way verification mechanism is used to calculate the actual length of the electric cylinder, including calculating the positive vector length from the moving hinge point AHP to the stationary hinge point OHP. and the length of the reverse vector from the static hinge point OHP to the moving hinge point AHP ;

[0028] S222, Calculate consistency error check: If the absolute difference between the length of the forward vector and the length of the reverse vector exceeds the consistency error threshold, it is determined to be a calculation failure, the error log is recorded and the fault handling procedure is started;

[0029] S223, Dynamic Safety Threshold Adaptive Adjustment: This feature dynamically adjusts the safety threshold for the link length based on the current acceleration state of the parallel robot platform, including the safety threshold for the maximum dynamically adjusted link length. and the safety threshold for the minimum dynamic adjustment lever length .

[0030] Furthermore, for the risk sources of pole length detection, if there is an interference risk, a graded risk management operation will be performed on the parallel robot platform at the response execution layer, specifically including:

[0031] Real-time comparison of the actual length of the electric cylinder Safety threshold for maximum dynamic adjustment lever length and the safety threshold for the minimum dynamic adjustment lever length :

[0032] S321: Normal state: It operates normally and does not trigger any security responses;

[0033] S322: Warning Status: or At this point, the pole length is close to the safety boundary, and a warning signal is issued to alert the user.

[0034] S323: Interference state: or If the pole length exceeds the safe range, the system will immediately trigger the corresponding safety response mechanism.

[0035] Furthermore, for singularity detection risk sources, the core detection layer uses the aforementioned basic data to determine whether there is interference risk among multiple risk sources of the parallel robot platform, specifically including:

[0036] S231, Precise Construction of the Jacobian Matrix: Based on the current pose state of the parallel robot platform, construct a complete 6×6 Jacobian matrix describing the kinematic relationships of the parallel robot. Each element in the matrix Indicates the first The speed of the electric cylinder joint affects the first Sensitivity to platform degrees of freedom, which include three translational degrees of freedom and three rotational degrees of freedom;

[0037] S232, Precise Calculation of Determinant and Numerical Stability Handling: The determinant value of the Jacobian matrix, i.e., the singularity test value, is calculated using a numerically stable LU decomposition algorithm. The full permutation formula was used for verification.

[0038] Furthermore, regarding the singularity detection risk source, if there is an interference risk, a graded risk management operation is performed on the parallel robot platform at the response execution layer, specifically including:

[0039] Accurate multi-level singularity assessment: Setting three levels of singularity assessment thresholds to achieve tiered early warning.

[0040] S331: Mild Singularity Warning: When When the condition is determined to be mildly singular, the parallel robot platform is still working normally, but its control performance begins to decline. Warning information is recorded for reference.

[0041] S332: Moderate Singularity Warning: When When the condition is determined to be a moderate singularity state, the control performance of the parallel robot platform deteriorates significantly, automatically reducing the movement speed and prompting an adjustment of the movement trajectory.

[0042] S333: Severe Singularity Warning: When When the condition is determined to be a severe singularity state, the parallel robot platform loses some degrees of freedom control capability and immediately executes the emergency shutdown procedure to ensure equipment safety.

[0043] Furthermore, for OBB collision detection risk sources, the core detection layer uses the aforementioned basic data to determine whether there is interference risk among multiple risk sources of the parallel robot platform, specifically including:

[0044] S241, OBB model accurate construction: For each detected object, construct the mathematical description of the OBB bounding box, including the center point coordinates of the center position of the OBB bounding box in the world coordinate system, the half-axis length vector representing the half length of the OBB bounding box in the three principal axis directions, and the local coordinate system orthogonal basis.

[0045] S242, Dynamic OBB Real-time Update Mechanism: For moving objects, including the electric cylinder, the OBB direction vector is updated in real time. This corresponds to an orthogonal basis of the local coordinate system;

[0046] S243: Perform collision detection, checking 15 potential separation axes, i.e., detection axes, including 3 body axes, 3 target axes, and 9 cross axes. The body axes are the 3 axes of the collision detection object A itself, and the target axes are the 3 axes of the target detection object B. Each body axis and each target axis will generate a cross vector, corresponding to 9 cross axes. And after performing a validity judgment on the cross product vector in the cross axes, the cross axis detection is performed.

[0047] S244, Precise Determination of Projection Separation: For each valid separation axis, calculate the separation distance; if any separation axis satisfies the separation distance... If the two OBBs separate, then they are considered to have separated; otherwise, a collision has occurred.

[0048] Furthermore, for OBB collision detection risk sources, if there is an interference risk, a graded risk management operation is performed on the parallel robot platform at the response execution layer, specifically including:

[0049] Safe distance threshold setting: To provide early warning, a safe distance threshold is set. When the minimum separation distance is less than the safe distance threshold, an early warning is triggered.

[0050] The beneficial effects of this invention are:

[0051] In an exemplary embodiment of the present invention, significant breakthroughs have been achieved in terms of comprehensiveness, accuracy, real-time performance, adaptability, and integration. This method, for the first time, integrates four major risk sources—Hooke's hinge, rod length, singularity, and OBB collision—into a unified framework, realizing full-dimensional safety monitoring from the internal workings of the mechanical body to the external environment. The detection accuracy reaches ±0.01°, and the response time is ≤1ms, meeting high-speed real-time requirements. The system possesses strong adaptability and scalability to dynamic working conditions, and through multi-module collaborative decision-making, effectively avoids response conflicts and decision delays, providing a systematic solution for the safe, efficient, and reliable operation of high-end parallel robots. Attached Figure Description

[0052] Figure 1 A flowchart of a full-dimensional self-interference detection method for a parallel robot platform provided as an exemplary embodiment of the present invention. Detailed Implementation

[0053] The technical solution of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Furthermore, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0054] See Figure 1 , Figure 1 A flowchart of a full-dimensional self-interference detection method for a parallel robot platform provided by an exemplary embodiment of the present invention is shown, including the following steps:

[0055] The data acquisition layer acquires basic data including six-degree-of-freedom pose information, electric cylinder length, and motion state information. The core detection layer uses the basic data to determine whether there is interference risk from multiple risk sources in the parallel robot platform. If there is interference risk, the response execution layer performs graded risk control operations on the parallel robot platform. Otherwise, the data acquisition and detection are continuously performed in a loop.

[0056] The parallel robot platform has multiple risk sources, including Hooke's hinge detection risk source, rod length detection risk source, singularity detection risk source, and OBB collision detection risk source.

[0057] Specifically, this exemplary embodiment achieves significant breakthroughs in comprehensiveness, accuracy, real-time performance, adaptability, and integration. This method, for the first time, integrates four major risk sources—Hooke's hinge, rod length, singularity, and OBB collision—into a unified framework, enabling full-dimensional safety monitoring from the internal workings of the mechanical body to the external environment. The detection accuracy reaches ±0.01°, and the response time is ≤1ms, meeting high-speed real-time requirements. The system possesses strong adaptability and scalability to dynamic operating conditions, and through multi-module collaborative decision-making, effectively avoids response conflicts and decision delays, providing a systematic solution for the safe, efficient, and reliable operation of high-end parallel robots.

[0058] The following content will elaborate on the preferred implementation methods for each content:

[0059] More preferably, in an exemplary embodiment, for the detection of risk sources using the Hooke's hinge, the core detection layer uses the basic data to determine whether there is interference risk among the multiple risk sources of the parallel robot platform, specifically including:

[0060] S211, Precise Calculation of Spatial Vectors: Based on the current motion pose of the parallel robot platform, the spatial vector of each Hooke hinge is calculated using an inverse kinematics algorithm. ; where the spatial vector The direction vector pointing from the lower hinge point (static hinge point) to the upper hinge point (moving hinge point) is expressed mathematically as follows:

[0061] ;

[0062] in, and These represent the coordinate positions of the upper hinge connection point and the lower hinge connection point in three-dimensional space, respectively. Represent spatial vectors respectively The x-coordinate, y-coordinate, and z-coordinate positions in three-dimensional space;

[0063] S212, Precise Dual-Axis Angle Calculation: Based on the spatial vector calculated in step S211 The biaxial angle of each Hooke hinge is calculated using a spatial vector precise decoupling method. The specific calculation process is as follows: First, the rotation angle about the Y-axis, which reflects the degree of inclination of the Hooke hinge in the vertical plane, is calculated. :

[0064] ;

[0065] Then, the rotation angle about the X-axis, which reflects the degree of rotation of the Hooke hinge in the horizontal plane, is calculated. :

[0066] ;

[0067] S213, Asymmetric constraint strategy detection: Interference detection is performed using an asymmetric constraint strategy based on the numbering characteristics of the Hooke's hinge, specifically including:

[0068] Odd-numbered hinge point constraint: For hinge points with odd numbers, a more lenient first constraint range is applied. ,in This is the maximum permissible angle (usually set to 25°). The minimum safe angle threshold;

[0069] Even-numbered hinge point constraint: For hinge points with even numbers, a second constraint range that is symmetrical but in the opposite direction is applied. To accommodate installation tolerances and load distribution;

[0070] All hinge points share a common constraint: all hinge points in the Y-axis direction adopt a uniform symmetric constraint, i.e., the third constraint range. ,in This represents the maximum permissible angle in the Y-axis direction.

[0071] Specifically, this exemplary embodiment discloses the detection of Hooke's hinge risk, with the following steps: First, precise spatial vector calculation is performed to obtain the precise direction vector of each hinge through inverse kinematics; then, precise decoupling of the two-axis angles is performed, independently calculating the X and Y axis rotation angles reflecting the degree of inclination of the hinge in the vertical and horizontal planes; finally, an innovative asymmetric constraint strategy is adopted, applying differentiated angle constraint thresholds based on the hinge number (odd / even) and installation characteristics, while ensuring uniform Y-axis constraints. This effectively overcomes the misjudgment problem caused by installation tolerances and uneven loads, significantly improving the accuracy of interference detection and system reliability.

[0072] Additionally, it should be noted that,

[0073] (1) In the angle calculation process of step S212, the arctangent function is used to ensure the continuity of the angle value, and by multiplying by Converting radians to angles facilitates engineering applications.

[0074] (2) Meanwhile, in step S213, according to the swing range limitation of the Hooke hinge in actual situation, asymmetry means that the reachable angle in the positive direction of the X-axis (+X) is inconsistent with the reachable angle in the negative direction of the X-axis (-X); the same applies to the Y-axis.

[0075] (3) Pairing Relationship: In a well-designed symmetrical parallel mechanism, hinges usually work in pairs. The most common case is that hinges numbered odd (1, 3, 5) and hinges numbered even (2, 4, 6) are spatially opposite each other or work together to form a stable motion chain or force unit. Odd-numbered hinge points refer to a group of hinges (e.g., numbered 1, 3, 5) that are in a specific phase (e.g., 0°, 120°, 240°) in the symmetrical layout of the mechanism; even-numbered hinge points refer to another group of hinges (e.g., numbered 2, 4, 6) that are paired with odd-numbered hinge points.

[0076] (4) The reason why misjudgment caused by installation tolerances and uneven loads can be overcome is that a dynamic theoretical benchmark is established by inverse kinematics, and the relative deviation between the actual value and the theoretical value is detected, rather than the absolute position; and asymmetric constraint thresholds are implemented for odd / even numbered hinges. This effectively filters out the initial pose differences caused by tolerances and focuses on abnormal and unexpected changes.

[0077] More preferably, in an exemplary embodiment, for the detection risk source of the Hooke's hinge, if there is an interference risk, a graded risk management operation is performed on the parallel robot platform at the response execution layer, specifically including:

[0078] The dual-axis angles of each Hooke hinge are monitored in real time. When any angle exceeds the preset constraint range, the corresponding safety response mechanism is immediately triggered.

[0079] S311: Level 1 warning: When the dual-axis angle approaches the preset proportion of the constraint range (such as reaching 90% of the constraint value), the system issues a warning signal to alert the user.

[0080] S312: Secondary speed reduction: When the dual-axis angle exceeds the constraint range but does not reach the danger threshold, the movement speed of the parallel robot platform is automatically reduced to a safe level;

[0081] S313: Level 3 Emergency Stop: When the dual-axis angles significantly exceed the constraint range or a rapid change trend is detected, an emergency stop procedure is immediately executed.

[0082] Specifically, in this exemplary embodiment, the core advantages of this hierarchical control mechanism are: (1) Precise response: different response levels are matched according to the severity of the risk, avoiding unnecessary shutdowns caused by a one-size-fits-all emergency stop. (2) Dynamic adaptation: from early warning to deceleration to emergency stop, a dynamically adjusted closed-loop safety loop is formed. (3) Ensuring continuity: through early warning and deceleration, buffer and adjustment time are provided for the system and operators, which helps to maintain the production cycle and improve the overall availability and reliability of the system.

[0083] Furthermore, regarding the detection of risk sources using the Hooke's hinge, this exemplary embodiment forms a complete and closed-loop processing chain from the core detection layer (S211-S213) to the response execution layer (S311-S313), fully demonstrating its high system integration and collaborative work characteristics, and effectively avoiding decision delays and response conflicts.

[0084] In a preferred exemplary embodiment, a hysteresis comparison algorithm is used in the interference judgment process to avoid frequent switching of response states near the boundary, ensuring the stability of system operation. For example: Level 1 warning: The trigger threshold may be 90% of the constraint value, but the threshold for releasing the warning will be set to 85%. The warning state will only be released if the angle continuously falls below 85%. Level 2 deceleration and Level 3 emergency stop: Similarly, corresponding entry and exit hysteresis bands are set.

[0085] The introduction of the hysteresis comparison algorithm brings the following core advantages: (1) Elimination of state oscillation: It effectively filters out interference caused by sensor noise or small system jitter, preventing the system from rapidly and repeatedly jumping between normal and warning, warning and deceleration states. (2) Improvement of decision stability: It ensures that the system's response state change is clear and firm, and only when the risk trend changes clearly and continuously will the state switch be triggered. (3) Enhancement of system robustness: It makes the entire safety control system more stable and predictable under dynamic and complex actual working conditions.

[0086] More preferably, in an exemplary embodiment, for the risk source of pole length detection, the core detection layer uses the basic data to determine whether there is interference risk among the multiple risk sources of the parallel robot platform, specifically including:

[0087] S221, Two-way Length Precision Verification: To ensure the accuracy of the rod length calculation, a two-way verification mechanism is used to calculate the actual length of the electric cylinder. First, the positive vector length from the moving hinge point AHP to the stationary hinge point OHP is calculated. :

[0088] ;

[0089] In the formula, This represents the x-coordinate position of the moving hinge point AHP in three-dimensional space. This represents the x-coordinate position of the static hinge point OHP in three-dimensional space. This represents the y-coordinate position of the moving hinge point AHP in three-dimensional space. This represents the y-coordinate position of the static hinge point OHP in three-dimensional space. This represents the z-coordinate position of the moving hinge point AHP in three-dimensional space. This represents the z-coordinate position of the static hinge point OHP in three-dimensional space.

[0090] Then calculate the length of the reverse vector from the static hinge point OHP to the moving hinge point AHP. :

[0091] ;

[0092] In theory, and They should be completely equal, but numerical errors may exist in actual calculations;

[0093] S222, Calculate the consistency error check: Calculate the absolute difference between the lengths of the forward vector and the reverse vector.

[0094] ;

[0095] Set the consistency error threshold to ,if If the error is detected, it is determined to be a calculation failure, the error log is recorded, and the fault handling procedure is started.

[0096] S223, Dynamic Safety Threshold Adaptive Adjustment: The safety threshold for the link length is dynamically adjusted based on the current acceleration state of the parallel robot platform.

[0097] ;

[0098] in, The safety threshold for the maximum dynamic adjustment rod length, This is the static maximum safety bar length. This is the current acceleration value (unit: m / s²). This is the acceleration compensation coefficient;

[0099] ;

[0100] The minimum safe threshold for the dynamic adjustment rod length (unit: mm). This is the static minimum safety rod length (unit: mm), which is the lower limit of the rod length when the system is stationary or at low speed. This is the current acceleration value (unit: m / s²). This is the acceleration compensation coefficient (consistent with the coefficient in the safety threshold formula for the maximum dynamic adjustment rod length).

[0101] Specifically, in this exemplary embodiment, to avoid potential numerical directionality errors in single-vector calculations, steps S221 and S222 employ a two-way verification mechanism: In step S21, the forward vector length from the moving hinge point to the stationary hinge point and the reverse vector length from the stationary hinge point to the moving hinge point are calculated simultaneously. Theoretically, these two should be absolutely equal, laying the foundation for subsequent consistency checks. Then, in step S222, the absolute difference between the forward and reverse lengths is calculated and compared with a preset consistency error threshold. If the difference exceeds the threshold, it is determined to be a numerical calculation fault or data anomaly, and the system will record an error log and initiate a fault handling procedure. This step is a crucial quality control step, ensuring the validity of the input data.

[0102] It should be noted that the "" mentioned in step S221 and They should be completely equal, but numerical errors may exist in actual calculations. The reasons for this include: (1) Precision limitations of floating-point representation. Computers use a finite number of bits (such as 32-bit or 64-bit binary numbers) to represent real numbers, which leads to small rounding errors in most numerical values. (2) Differences in the order and path of operations. Different calculation formulas or orders of operations, even if mathematically equivalent, will result in differences in the final results due to different ways of accumulating rounding errors in intermediate steps. (3) Precision loss in mathematical operations. Complex mathematical operations (such as squaring, square rooting, trigonometric functions) will introduce new and unavoidable approximation errors during the calculation process.

[0103] In step S223, considering the dynamic effects during high-speed motion, an adaptive threshold adjustment strategy for acceleration is adopted. When the platform is in a high-speed acceleration or deceleration state, the actual safety bar length threshold will shrink accordingly, reserving a safety margin for dynamic effects. This adaptive adjustment strategy significantly reduces the false alarm rate under high-speed operating conditions.

[0104] Therefore, the following advantages are available for rod length detection: (1) Computational robustness: Two-way verification and consistency checks constitute a double insurance, effectively identifying and isolating abnormal data caused by floating-point operations, sensor jumps, etc., and improving detection reliability from the source. (2) Dynamic safety: The adaptive adjustment strategy of dynamic safety threshold is the highlight of this method. It enables the system to sense its own motion state and automatically adopt a more conservative safety boundary under high-speed and high-acceleration conditions, significantly reducing the risk of missed or delayed reporting due to ignoring dynamic effects, and is especially suitable for dynamic scenarios such as high-speed handling and simulated vibration. (3) Engineering applicability: Through parameters This strategy can be easily calibrated according to the dynamic characteristics of different robots, demonstrating good configurability and engineering adaptability.

[0105] More preferably, in an exemplary embodiment, for the risk source of pole length detection, if there is an interference risk, a graded risk management operation is performed on the parallel robot platform at the response execution layer, specifically including:

[0106] Real-time comparison of the actual length of the electric cylinder Safety threshold for maximum dynamic adjustment lever length and the safety threshold for the minimum dynamic adjustment lever length :

[0107] S321: Normal state: It operates normally and does not trigger any security responses;

[0108] S322: Warning Status: or At this point, the pole length is close to the safety boundary, and a warning signal is issued to alert the user.

[0109] S323: Interference state: or If the pole length exceeds the safe range, the system will immediately trigger the corresponding safety response mechanism.

[0110] Specifically, in this exemplary embodiment, this response strategy is closely integrated with the dynamic threshold of the core detection layer, demonstrating the following advantages: (1) Seamless integration with dynamic threshold: The response judgment is directly based on the dynamically adjusted safety threshold, ensuring that the response mechanism remains accurate and effective under dynamic conditions such as high-speed acceleration and deceleration, and realizing closed-loop adaptive detection and response. (2) Clear and reasonable state division: By setting the warning zone (0.95, 1.05), a buffer zone is established between complete safety and necessary intervention. This avoids frequent false alarms near the safety boundary (the effect is better when combined with the hysteresis comparison algorithm), and provides a path for gradual risk escalation, improving the smoothness and operational friendliness of the system. (3) Differentiated response measures: Different response actions are taken for warning and intervention states (prompt vs. forced intervention), which reflects the core idea of ​​hierarchical control, ensuring safety while minimizing the interruption of normal work processes.

[0111] More preferably, in an exemplary embodiment, for singularity detection risk sources, the core detection layer uses the basic data to determine whether there is interference risk among multiple risk sources of the parallel robot platform, specifically including:

[0112] S231, Precise Construction of the Jacobian Matrix: Based on the current pose state of the parallel robot platform, construct a complete 6×6 Jacobian matrix describing the kinematic relationships of the parallel robot. Each element in the matrix Indicates the first The speed of the electric cylinder joint affects the first Sensitivity to platform degrees of freedom, which include three translational degrees of freedom and three rotational degrees of freedom:

[0113] For each electric cylinder , Calculate the unit vector That is, the direction vector from the static hinge point to the moving hinge point, and the position vector of the moving hinge point relative to the center of mass of the platform. The Jacobian matrix of the first... The row is determined by the following formula:

[0114] ;

[0115] in, The influence of translational degrees of freedom The influence of the rotational degree of freedom is considered; the Jacobian matrix constructed in this way fully describes the kinematic properties of the platform.

[0116] S232, Precise Calculation of Determinant and Numerical Stability Handling: The determinant value of the Jacobian matrix, i.e., the singularity test value, is calculated using a numerically stable LU decomposition algorithm. :

[0117] ;

[0118] in, and These are the diagonal elements of the lower triangular matrix and the upper triangular matrix obtained by performing LU decomposition on the Jacobian matrix J, respectively. Let J represent the determinant of the Jacobian matrix J;

[0119] Simultaneously, the full permutation formula was used for verification:

[0120] ;

[0121] in It is a 6th order symmetric group. For arrangement, The symbols for arrangement, This indicates that the i-th row and i-th column of the Jacobian matrix J are taken. The elements of the column; the difference between the results of the two calculation methods should be less than [value missing]. Otherwise, recalculate.

[0122] Specifically, in this exemplary embodiment, the risk analysis process for singularity detection risk sources is as follows:

[0123] First, in step S231, the Jacobian matrix is ​​precisely constructed to achieve modeling completeness: a complete 6×6 Jacobian matrix J is rigorously constructed to describe the mapping relationship between all 6 degrees of freedom (3 translations + 3 rotations) and the velocities of the 6 electric cylinders; and the physical meaning is clear, with the i-th row of the matrix corresponding to the contribution of the i-th electric cylinder, consisting of two items: the direction vector from the static hinge point to the moving hinge point. This directly reflects the influence of platform translation on the change in the length of the electric cylinder; the position vector of the moving hinge point relative to the platform's center of mass. and cross product This reflects the effect of platform rotation on the change in the length of the electric cylinder. This construction method, based on the velocity-level kinematics of parallel robots, is the gold standard for singularity analysis.

[0124] Then, in step S232, the determinant is calculated precisely and its numerical stability is processed, including: (1) Main calculation method (efficient and stable): The determinant value det(J) is calculated using the numerically stable LU decomposition algorithm. The determinant is equal to the product of all diagonal elements of the lower triangular matrix L and the upper triangular matrix U. LU decomposition can effectively handle ill-conditioned matrices and is the preferred method for industrial calculation. (2) Verification method (theoretical benchmark): The Leibniz formula based on full permutations is used for calculation. This method is clearly defined and can be used as a theoretical benchmark for cross-validation. (3) Consistency check (reliability guarantee): The absolute difference between the results of the two independent methods must be less than the preset tolerance. If the tolerance is exceeded, a recalculation or fault flag is triggered. This dual verification mechanism is the essence of this step. It greatly eliminates the numerical errors that may occur in extreme configurations of a single algorithm and ensures the absolute reliability of det(J) as a singularity criterion.

[0125] It should be noted that in step S232:

[0126] (1) For a Jacobian matrix J, its determinant value is denoted as det(J). When judging singularity: when det(J)=0, the Jacobian matrix J is singular (non-invertible). At this time, the robot arm is in a singular configuration, which means that at this position, the end effector of the robot arm will lose its motion capability in some directions or the joint speed will tend to infinity.

[0127] (2) For LU decomposition, performing LU decomposition on the Jacobian matrix J in the control of robotic arms or parallel robots is a commonly used and numerically stable calculation method for singularity judgment. Specifically: performing LU decomposition on the Jacobian matrix J describing the robot's velocity mapping relationship means decomposing it into the product of a lower triangular matrix L and an upper triangular matrix U (J=L×U). After decomposition, the numerical calculation of the determinant value det(J) of matrix J can be simplified to the product of all diagonal elements of L and U (i.e., This is because the determinant of a triangular matrix is ​​equal to the product of its diagonal elements. Directly calculating the determinant can be numerically unstable, while LU decomposition is a standard algorithm with high numerical stability and excellent efficiency. By calculating whether this product value is close to zero (less than...),... This allows for stable and accurate determination of whether the robotic arm is in a singular configuration, which is crucial for avoiding joint speed loss in real-time control.

[0128] (3) In the formula for all permutations, This refers to the mathematical set of all possible permutations of 6 elements (corresponding to 6 joints or 6 rows / columns). The calculation requires traversing all permutations in this set (6! = 720 in total). It is a specific permutation taken from S6, for example, σ=(2,1,3,4,5,6). In the calculation, it is generated one by one by a traversal algorithm (such as recursion, iteration). The number of inversions is calculated based on the number of inversions in the permutation. If the number of inversions is even, then... If it is an odd number, then Used to determine the sign of the term in the summation. This is the key to the determinant calculation formula: • J is a 6×6 Jacobian matrix, where each element... Calculated from the robot's kinematic model (using position, joint parameters, etc.). For a given arrangement , This means taking the i-th row and the first row of matrix J. The elements of the column. The product of the six elements taken from all rows according to this rule constitutes one term in the summation.

[0129] (4) For recalculation, a controlled perturbation of the computation path can be introduced. Specifically, a small, random regularization perturbation is applied to the elements of the Jacobian matrix J, for example, calculating J = J + ε * I, where I is the identity matrix and ε is a very small scalar (e.g., 10). -10 Then, det(J) is calculated using both the formula method and the LU decomposition method. This is equivalent to moving the problem away from a cliff that might be near a numerical singularity, and calculating it from a neighboring point with a better condition number. The difference between the two calculation results should become minimal and acceptable. Compare the two calculation results of det(J) with the difference before the perturbation. If the difference is significantly reduced after perturbation, it inversely confirms that the original matrix J itself is in a numerically sensitive region.

[0130] Therefore, in this exemplary embodiment, a complete Jacobian matrix is ​​rigorously constructed, ensuring the theoretical completeness of singular configuration detection without any risk of omission. Simultaneously, it exhibits excellent numerical robustness, a wise choice of master algorithm, and superior numerical stability and computational efficiency compared to direct computation through LU decomposition. Furthermore, a robust verification mechanism is implemented, introducing a permutation-based verification formula to form a double insurance of master computation and verification. This is not merely simple redundant computation, but rather cross-validation utilizing different mathematical principles, effectively capturing numerical instability issues caused by the extreme difference in matrix condition numbers (approaching singularity).

[0131] More preferably, in an exemplary embodiment, for singularity detection risk sources, the step of performing graded risk control operations on the parallel robot platform at the response execution layer if there is interference risk specifically includes:

[0132] Accurate multi-level singularity assessment: Setting three levels of singularity assessment thresholds to achieve tiered early warning.

[0133] S331: Mild Singularity Warning: When When the condition is determined to be mildly singular, the parallel robot platform is still working normally, but its control performance begins to decline. Warning information is recorded for reference.

[0134] S332: Moderate Singularity Warning: When When the condition is determined to be a moderate singularity state, the control performance of the parallel robot platform deteriorates significantly, automatically reducing the movement speed and prompting an adjustment of the movement trajectory.

[0135] S333: Severe Singularity Warning: When When the condition is determined to be a severe singularity state, the parallel robot platform loses some degrees of freedom control capability and immediately executes the emergency shutdown procedure to ensure equipment safety.

[0136] Specifically, in this exemplary embodiment, based on the Jacobian matrix determinant value |det(J)| accurately calculated by the core detection layer, a three-level judgment threshold is set in the response execution layer to implement progressive risk intervention:

[0137] In step S331, there is a mild singularity warning (performance monitoring state). At this time, the platform begins to approach the singular region, the condition number of the Jacobian matrix increases, and the control performance (such as accuracy and force transmission efficiency) begins to decline slightly, but is still within the controllable range. The system only records the warning log and may provide status indications without interfering with the current movement. This stage mainly provides early performance degradation information to the operator or the upper-level system for trend analysis and preventive planning.

[0138] In step S332, a moderate singularity warning (active protection state) is issued. The platform is very close to a singular configuration, the control performance has deteriorated significantly, the joint force / velocity may increase sharply, and the risk of continuing to move along the original trajectory is very high. The system automatically triggers protective measures, which usually include: reducing the end-effector velocity to reduce joint demand; prompting or requesting adjustment of the motion trajectory to move away from the singular direction. At this stage, the risk is actively reduced while ensuring that the system does not stop, providing an opportunity for a soft landing.

[0139] Step S333 is a severe singularity warning (safety protection state). The platform has entered or is about to enter a singular configuration, the Jacobian matrix is ​​rank deficient, and the robot loses control in one or more directions, potentially leading to speed loss, motor overload, or other dangers. The system immediately executes the highest level of safety response—the emergency stop procedure. This is the final barrier to prevent mechanical damage and ensure absolute safety.

[0140] Therefore, in this exemplary embodiment: (1) Quantify risks and intervene in advance: Two-level buffer zones are established between complete normality and complete loss of control through two thresholds. This achieves a smooth transition from performance monitoring to proactive avoidance and then to emergency protection, eliminating accidents in their infancy. (2) Balance safety and efficiency: Mild warnings do not interrupt production, moderate warnings reduce speed to gain opportunities for online adjustments, and only severe warnings force shutdowns. This minimizes unnecessary production interruptions and improves the overall utilization rate of equipment. (3) Deep integration with high-precision detection: The effectiveness of this strategy depends entirely on the high-precision and high-reliability |det(J)| value provided by the core detection layer. It is the rigor of steps S231-S232 that ensures the accuracy of threshold judgment here, avoiding accidental shutdowns caused by misjudgments or safety accidents caused by missed judgments.

[0141] In a preferred exemplary embodiment, an adaptive early warning response mechanism is employed, implementing a graded response strategy based on the degree of singularity: Moderate response: The system automatically limits the maximum movement speed to 50% of the normal value, and simultaneously activates a trajectory optimization algorithm to find alternative paths to avoid the singular region. Severe response: The system immediately executes an emergency stop, all electric cylinders stop moving at maximum safe deceleration, and an audible and visual alarm signal is issued.

[0142] More preferably, in an exemplary embodiment, for OBB collision detection risk sources, the core detection layer uses the basic data to determine whether there is interference risk among multiple risk sources of the parallel robot platform, specifically including:

[0143] S241, Precise OBB Model Construction: For each detected object (obtained from the user's 3D model or entity data), construct a mathematical description of the OBB bounding box:

[0144] Center point coordinates: The center point coordinates represent the center position of the OBB bounding box in the world coordinate system. These represent the x-coordinate position, y-coordinate position, and z-coordinate position of the center point in three-dimensional space, respectively.

[0145] Semi-axis length vector: The half-axis length vector represents the half-length of the OBB bounding box along the three principal axes (the X, Y, and Z axes of the bounding box's own coordinate system). These represent the half-lengths of the OBB bounding box along the three local coordinate axes, respectively.

[0146] Local coordinate system orthogonal basis: Let X, Y, and Z represent the three unit direction vectors of the local coordinate system of the OBB bounding box (i.e., the directions of the X, Y, and Z axes of the local coordinate system, the unit axes), also known as basis vectors or principal axis vectors, satisfying the orthogonality normalization condition:

[0147] ;

[0148] In the formula, and Let represent the i-th unit direction vector and the j-th unit direction vector in the local coordinate system, respectively. The symbol represents the Kronecker delta, and i and j represent the vector subscripts (1, 2, and 3 correspond to the X-axis, Y-axis, and Z-axis of the local coordinate system, respectively).

[0149] Here, the center point coordinates represent the position of the geometric center of the bounding box in the world coordinate system; the semi-axis length vectors represent the half-lengths along the three principal axes, determining the size of the bounding box; and the local coordinate system orthogonal basis consists of three mutually orthogonal unit vectors, defining the orientation of the bounding box. This orthogonal normalization condition is the foundation for the correctness of subsequent projection calculations.

[0150] S242, Dynamic OBB Real-time Update Mechanism: For moving objects, including the electric cylinder, the OBB direction vector is updated in real time. :

[0151] ;

[0152] in, Let be the coordinates of the upper hinge point. Let these be the coordinates of the lower hinge point; each The corresponding components within the OBB enclosure include Three directions.

[0153] S243: Perform collision detection, checking 15 potential separation axes (detection axes), including 3 body axes, 3 target axes, and 9 cross axes. The body axes are the three axes of the collision detection object A itself, and the target axes are the three axes of the target detection object B. The detection here sequentially checks the situation of a certain object relative to all other potentially interfering objects, thus defining the concepts of body and target. These 6 axes (3 body axes and 3 target axes) generate a cross vector for every two axes, corresponding to a total of 9 cross axes.

[0154] S2431, 3 body axis detection: for the local coordinate system axes of OBB A Calculate the body axis projection:

[0155] ;

[0156] Where i represents the index of the three body axes of A (i = 1, 2, 3, where 1, 2, 3 correspond to X, Y, Z respectively; therefore, for , Corresponding to ), corresponding to the subsequent projection radius; The distance vector from the center point. and These represent the center coordinates of A and B, respectively.

[0157] Let A be on the axis The projection radius on the plane, i is the corresponding 3 body axes of A, and k represents the sequence number of the three local coordinate system axes of A; Let represent the component of the semi-axis length vector of A along the k-axis. Let A be the k-th unit direction vector. Let A represent the i-th unit direction vector.

[0158] For B on the axis The projection radius on the plane, i is the corresponding 3 body axes of A, and k represents the sequence number of the three local coordinate system axes of B; This represents the component of the semi-axis length vector of B along the k-axis. Let B be the k-th unit direction vector. Let A represent the i-th unit direction vector.

[0159] S2432, 3 target axis detection: for the local coordinate system axes of OBBB Calculate the target axis projection:

[0160] ;

[0161] Where j represents the index of the three body axes of B (j=1, 2, 3, where 1, 2, 3 correspond to X, Y, Z respectively; therefore, for , Corresponding to );

[0162] Let A be on the axis The projection radius on the coordinate system is given by: j represents the three body axes of B, and k represents the index of the three local coordinate system axes of B. Let represent the component of the semi-axis length vector of A along the k-axis. Let A be the k-th unit direction vector. Let B be the j-th unit direction vector.

[0163] For B in The projection radius on the axis, j corresponds to the three body axes of B, and k represents the index of the three local coordinate system axes of B. This represents the component of the semi-axis length vector of B along the k-axis. Let B be the k-th unit direction vector. Let B be the j-th unit direction vector.

[0164] S2433, Intelligent Zero Vector Removal: For the cross product vector of the cross axis To determine validity:

[0165] ;

[0166] in, Threshold If the requirements are met, the detection axis is skipped to avoid invalid detection caused by numerical calculation errors;

[0167] S2434, 9-cross axis detection: For all cross axes that meet the requirements of S2433. ( ), calculate the cross-axis projection:

[0168] ;

[0169] in, Normalized cross-axis vector;

[0170] S244, Precise Determination of Projection Separation: For each valid separation axis, determine the projection of the body axis of the valid separation axis. Target axis projection Cross axis projection The separation distance represented by the distance, if there exists a separation axis whose separation distance satisfies If the two OBBs separate, they will separate; otherwise, a collision will occur.

[0171] Specifically, in this exemplary embodiment, in step S241, the OBB model is accurately constructed, and an OBB that closely fits the geometric contour of each component to be inspected is established; while in step S242, the dynamic OBB real-time update mechanism is implemented. For moving objects such as electric cylinders, the direction (especially the axial direction) of the OBB changes in real time with the movement. This step dynamically calculates and updates the OBB direction vector based on the coordinates AHP and OHP of its upper and lower hinge points. Ensuring that the OBB model always maintains the same posture as the physical component is a prerequisite for achieving real-time collision detection.

[0172] In step S243, fast collision detection based on the separating axis theorem requires checking 15 potential separating axes (L) to determine whether two OBBs (A and B) intersect:

[0173] In step S2431, the three body axes are detected: the projections of the three body axes of OBB A are checked. .

[0174] In step S2432, three target axes are detected: the projections of the three body axes of OBB itself are checked. .

[0175] In step S2433, zero vector intelligent removal (algorithm optimization) is performed in step S244, which means adding a pre-filtering step before performing the above 9 cross-axis detections. For the cross axes obtained by the cross product, the cross product vector of the cross axis is calculated. If its value is less than a very small threshold If the vector is approximately zero, it indicates that the two corresponding original axes are nearly parallel, and its cross product axis is invalid, so this detection can be skipped directly. This optimization significantly reduces unnecessary computation and improves the overall detection speed.

[0176] More specifically, the purpose of step S2433 is to ensure the numerical stability and robustness of the Separated Axis Theorem (SAT) in practical calculations. When any pair of principal axes of two OBBs... and When they are approximately parallel (the included angle is close to 0° or 180°), their cross product vector... The length is close to zero (theoretical value is 0, but in practice it is usually a very small value due to floating-point errors, preferably...). If such degenerate axes are not removed, subsequent normalization operations (n ​​= cross product vector / its length) will result in: division by zero errors or invalid values ​​such as NaN / Inf; the direction vector will be distorted due to noise, leading to unreliable projection judgments and a high risk of misjudgment. To avoid these problems, this step performs a length check on each potential cross product vector before calculating the 9 cross axes: if... If the axis is not parallel, the subsequent projection test of that axis is skipped. This elimination mechanism does not affect the correctness of the detection for the following reasons: (1) When the two axes are parallel, their separation information has been fully covered by the main axis of A or B (the first 6 axes), so there is no need to test the degenerate cross axis. (2) Skipping invalid axes can reduce invalid calculations (saving up to 9 projections) and significantly improve the numerical stability of the algorithm in dynamic scenes (such as electric cylinder movement).

[0177] In step S2434, the nine cross-axis detection involves checking the projections of the nine cross axes obtained from the pairwise cross products of axes A and B. (If removed from S2433, the quantity will be reduced accordingly).

[0178] Finally, in step S244, precise judgment of projection separation is performed. All valid detection axes are traversed, and if any axis is found to satisfy |T·L| > ( + If the condition is met, the two OBBs can be determined to be separated (no collision). If none of the 15 axes meet this condition, then a collision is determined.

[0179] This OBB collision detection scheme is a model of achieving high speed, real-time performance and high reliability in this method: (1) High computational efficiency: The OBB model, combined with the separating axis theorem, intelligently eliminates zero vectors, further optimizing the performance, enabling the algorithm to complete multiple pairs of detections in complex scenes within ≤1ms, meeting the real-time requirements of high-speed motion scenes. (2) Reliable detection accuracy: Compared with spheres or AABBs (axial bounding boxes), OBBs wrap around slender parts (such as electric cylinders) more tightly, reducing the "false alarm" space and improving detection accuracy. The complete detection of 15 axes mathematically guarantees the accuracy of the results. (3) Strong dynamic adaptability: By updating the OBB direction of moving objects in real time, the algorithm can accurately track the continuous motion of the robot and achieve accurate collision warning between consecutive frames. (4) Modularity and scalability: The detection module can work independently and is easy to integrate new objects to be detected. Its core parameters (such as OBB size and threshold) can be calibrated according to the specific robot model, reflecting good engineering adaptability and scalability.

[0180] More preferably, in an exemplary embodiment, for OBB collision detection risk sources, if there is an interference risk, a graded risk management operation is performed on the parallel robot platform at the response execution layer, specifically including:

[0181] Safe distance threshold setting: Set a safe distance threshold for early warning. The collision warning conditions are:

[0182]

[0183] That is, an early warning is triggered when the minimum separation distance is less than the safe distance threshold.

[0184] Specifically, in this exemplary embodiment, a safety distance threshold is introduced for tiered control to achieve proactive protection, extending the safety boundary outward from the geometric surface of the object by a certain distance. A virtual buffer layer. This allows the system to identify risks and issue early warnings before physical contact occurs, transforming passive collision detection into proactive proximity warnings, thus significantly improving safety.

[0185] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art can make other variations or modifications based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A method for full-dimensional self-interference detection of parallel robot platforms, characterized by: Includes the following steps: The data acquisition layer acquires basic data including six-degree-of-freedom pose information, electric cylinder length, and motion state information. The core detection layer uses the basic data to determine whether there is interference risk from multiple risk sources in the parallel robot platform. If there is interference risk, the response execution layer performs graded risk control operations on the parallel robot platform. Otherwise, the data acquisition and detection are continuously performed in a loop. Among them, the multiple risk sources of the parallel robot platform include Hooke's hinge detection risk source, rod length detection risk source, singularity detection risk source and OBB collision detection risk source; For the detection of risk sources in the Hooke's hinge, the core detection layer uses the aforementioned basic data to determine whether there is interference risk among the multiple risk sources of the parallel robot platform, specifically including: S211, Precise calculation of spatial vectors: Based on the current motion pose of the parallel robot platform, the spatial vector of each Hooke hinge is calculated using the inverse kinematics algorithm; S212, Precise Calculation of Biaxial Angles: Based on the spatial vector obtained in step S211, calculate the biaxial angles of each Hooke hinge, including the rotation angle about the Y-axis that reflects the degree of inclination of the Hooke hinge in the vertical plane. And the rotation angle about the X-axis, which reflects the degree of rotation of the Hooke hinge in the horizontal plane. ; S213, Asymmetric Constraint Strategy Detection: Based on the numbering characteristics of Hooke's hinges, an asymmetric constraint strategy is used for interference detection based on the constraint range, specifically including: Odd-numbered hinge point constraint: For hinge points with odd numbers, a more lenient first constraint range is applied. ,in For the maximum permissible angle, The minimum safe angle threshold; Even-numbered hinge point constraint: For hinge points with even numbers, a second constraint range that is symmetrical but in the opposite direction is applied. To accommodate installation tolerances and load distribution; All hinge points share a common constraint: all hinge points in the Y-axis direction adopt a uniform symmetric constraint, i.e., the third constraint range. ,in This represents the maximum permissible angle in the Y-axis direction.

2. The method for full-dimensional self-interference detection of a parallel robot platform according to claim 1, characterized in that: For the risk sources of Hooke's hinge detection, if there is interference risk, a graded risk management operation will be performed on the parallel robot platform at the response execution layer, specifically including: The dual-axis angles of each Hooke hinge are monitored in real time. When any angle exceeds the preset constraint range, the corresponding safety response mechanism is immediately triggered. S311: Level 1 warning: When the dual-axis angle approaches the preset ratio of the constraint range, the system issues a warning signal to alert the user; S312: Secondary speed reduction: When the dual-axis angle exceeds the constraint range but does not reach the danger threshold, the movement speed of the parallel robot platform is automatically reduced to a safe level; S313: Level 3 Emergency Stop: When the dual-axis angles significantly exceed the constraint range or a rapid change trend is detected, an emergency stop procedure is immediately executed.

3. The method for full-dimensional self-interference detection of a parallel robot platform according to claim 1, characterized in that: For the risk sources of pole length detection, the core detection layer uses the aforementioned basic data to determine whether there is interference risk among the multiple risk sources of the parallel robot platform, specifically including: S221, Two-way Length Precision Verification: To ensure the accuracy of rod length calculation, a two-way verification mechanism is used to calculate the actual length of the electric cylinder, including calculating the positive vector length from the moving hinge point AHP to the stationary hinge point OHP. and the length of the reverse vector from the static hinge point OHP to the moving hinge point AHP ; S222, Calculate consistency error check: If the absolute difference between the length of the forward vector and the length of the reverse vector exceeds the consistency error threshold, it is determined to be a calculation failure, the error log is recorded and the fault handling procedure is started; S223, Dynamic Safety Threshold Adaptive Adjustment: This feature dynamically adjusts the safety threshold for the link length based on the current acceleration state of the parallel robot platform, including the safety threshold for the maximum dynamically adjusted link length. and the safety threshold for the minimum dynamic adjustment lever length .

4. The method for full-dimensional self-interference detection of a parallel robot platform according to claim 3, characterized in that: For risks related to pole length detection, if there is an interference risk, a graded risk management operation will be performed on the parallel robot platform at the response execution layer, specifically including: Real-time comparison of the actual length of the electric cylinder Safety threshold for maximum dynamic adjustment lever length and the safety threshold for the minimum dynamic adjustment lever length : S321: Normal state: It operates normally and does not trigger any security responses; S322: Warning Status: or At this point, the pole length is close to the safety boundary, and a warning signal is issued to alert the user. S323: Interference state: or If the pole length exceeds the safe range, the system will immediately trigger the corresponding safety response mechanism.

5. The method for full-dimensional self-interference detection of a parallel robot platform according to claim 1, characterized in that: For singularity detection risk sources, the core detection layer uses the aforementioned basic data to determine whether there is interference risk among multiple risk sources of the parallel robot platform, specifically including: S231, Precise Construction of the Jacobian Matrix: Based on the current pose state of the parallel robot platform, construct a complete 6×6 Jacobian matrix describing the kinematic relationships of the parallel robot. Each element in the matrix Indicates the first The speed of the electric cylinder joint affects the first Sensitivity to platform degrees of freedom, which include three translational degrees of freedom and three rotational degrees of freedom; S232, Precise Calculation of Determinant and Numerical Stability Handling: The determinant value of the Jacobian matrix, i.e., the singularity test value, is calculated using a numerically stable LU decomposition algorithm. The full permutation formula was used for verification.

6. The method for full-dimensional self-interference detection of a parallel robot platform according to claim 5, characterized in that: For singularity detection risk sources, if there is interference risk, the parallel robot platform will be subject to graded risk control operations at the response execution layer, specifically including: Accurate multi-level singularity assessment: Setting three levels of singularity assessment thresholds to achieve tiered early warning. S331: Mild Singularity Warning: When When the condition is determined to be mildly singular, the parallel robot platform is still working normally, but its control performance begins to decline. Warning information is recorded for reference. S332: Moderate Singularity Warning: When When the condition is determined to be a moderate singularity state, the control performance of the parallel robot platform deteriorates significantly, automatically reducing the movement speed and prompting an adjustment of the movement trajectory. S333: Severe Singularity Warning: When When the condition is determined to be a severe singularity state, the parallel robot platform loses some degrees of freedom control capability and immediately executes the emergency shutdown procedure to ensure equipment safety.

7. The method for full-dimensional self-interference detection of a parallel robot platform according to claim 1, characterized in that: For OBB collision detection risk sources, the core detection layer uses the aforementioned basic data to determine whether there is interference risk among multiple risk sources of the parallel robot platform, specifically including: S241, OBB model accurate construction: For each detected object, construct the mathematical description of the OBB bounding box, including the center point coordinates of the center position of the OBB bounding box in the world coordinate system, the half-axis length vector representing the half length of the OBB bounding box in the three principal axis directions, and the local coordinate system orthogonal basis. S242, Dynamic OBB Real-time Update Mechanism: For moving objects, including the electric cylinder, the OBB direction vector is updated in real time. This corresponds to an orthogonal basis of the local coordinate system; S243: Perform collision detection, checking 15 potential separation axes, i.e., detection axes, including 3 body axes, 3 target axes, and 9 cross axes. The body axes are the 3 axes of the collision detection object A itself, and the target axes are the 3 axes of the target detection object B. Each body axis and each target axis will generate a cross vector, corresponding to 9 cross axes. And after performing a validity judgment on the cross product vector in the cross axes, the cross axis detection is performed. S244, Precise Determination of Projection Separation: For each valid separation axis, calculate the separation distance; if any separation axis satisfies the separation distance... If the two OBBs separate, then they are considered to have separated; otherwise, a collision has occurred.

8. The method for full-dimensional self-interference detection of a parallel robot platform according to claim 7, characterized in that: For OBB collision detection risk sources, if there is interference risk, a graded risk management operation will be performed on the parallel robot platform at the response execution layer, specifically including: Safe distance threshold setting: To provide early warning, a safe distance threshold is set. When the minimum separation distance is less than the safe distance threshold, an early warning is triggered.

Images