Multi-mechanical-arm dynamic time sequence anti-interference and trajectory smoothing control system

Abstract

The invention relates to the field of robots, and particularly discloses a multi-mechanical-arm dynamic time sequence anti-interference and trajectory smoothing control system which comprises a state monitoring and trajectory smoothing module, a dynamic time sequence planner, an energy consumption balancing optimizer and an interference detection and correction module. The state monitoring and trajectory smoothing module generates smooth trajectory data through a quintic polynomial interpolation algorithm; the dynamic time sequence planner adjusts the action time sequence of the mechanical arm in real time based on the task progress and the space occupation state; the energy consumption balance optimizer realizes balance optimization of energy consumption balance and operation efficiency through a decomposition-based multi-objective evolutionary algorithm; the interference detection and correction module generates an anti-interference correction instruction sensitive to energy consumption through an improved repulsive force field algorithm. According to the method, energy consumption balance optimization is fused into the whole process of multi-mechanical-arm cooperative control, the system energy efficiency is remarkably improved, the service life of equipment is remarkably prolonged while track smoothness and anti-interference safety are guaranteed, and the method is particularly suitable for intelligent manufacturing scenes with high-density layout.

Description

Technical Field

[0001] This invention relates to the field of robotics, and in particular to a dynamic timing anti-interference and trajectory smoothing control system for multiple robotic arms. Background Technology

[0002] With the continuous improvement of industrial automation, multi-robotic arm collaborative operation systems are increasingly widely used in intelligent manufacturing, assembly lines, and logistics sorting. However, existing technologies still face many technical bottlenecks in achieving efficient and safe multi-robotic arm collaborative control:

[0003] First, traditional trajectory planning methods struggle to balance smoothness and energy efficiency. Most current systems employ cubic spline interpolation or Bézier curves for trajectory planning, which, while ensuring positional continuity, cannot simultaneously meet the requirements for continuous and smooth velocity and acceleration, leading to shocks and vibrations during robotic arm operation. More importantly, existing trajectory planning methods rarely consider energy consumption factors and lack energy balancing mechanisms among multiple robotic arms, resulting in overall low system energy efficiency and shortened lifespans for some robotic arms due to prolonged high-load operation.

[0004] Secondly, the synergy between dynamic timing planning and anti-interference control is insufficient. Existing systems mostly employ fixed timing scheduling strategies or simple priority rules, failing to dynamically adjust the timing of robotic arm movements based on real-time task progress. In the event of unforeseen circumstances, this can easily lead to spatial interference between robotic arms or system stagnation. Although anti-interference control based on the artificial potential field method can avoid collisions, traditional repulsive force field models only consider distance factors and do not incorporate energy consumption constraints into the decision-making process, resulting in low energy efficiency of the avoidance path.

[0005] Furthermore, a systematic solution for multi-objective collaborative optimization is lacking. Current research largely focuses on optimizing single technical indicators, lacking comprehensive solutions that can simultaneously optimize trajectory smoothness, temporal efficiency, anti-interference security, and energy economy. For example, while temporal Petri net-based time-series planning methods can improve task scheduling efficiency, they do not consider energy consumption distribution; while model predictive control-based anti-interference strategies can ensure security, their high computational complexity makes it difficult to meet real-time requirements.

[0006] The aforementioned problems are particularly pronounced in high-density collaborative operation scenarios. When multiple robotic arms operate in parallel within a confined space, ensuring smooth, impact-free trajectories, achieving dynamic interference prevention, and optimizing system energy consumption place extremely high demands on the overall performance of the control system. Existing technologies often can only address one or two of these objectives, making it difficult to achieve collaborative optimization of multiple objectives.

[0007] Therefore, there is an urgent need to develop a multi-manipulator collaborative control system that can integrate trajectory smoothing control, dynamic timing planning, energy consumption balance optimization, and intelligent anti-interference to comprehensively improve the operating efficiency, safety, and economy of multi-manipulator systems. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of existing technologies and propose a multi-robotic arm dynamic timing anti-interference and trajectory smoothing control system.

[0009] To achieve the above objectives, the present invention adopts the following technical solution:

[0010] The multi-robotic arm dynamic timing anti-interference and trajectory smoothing control system includes four core processing modules connected sequentially via a high-speed data bus, specifically:

[0011] The status monitoring and trajectory smoothing module is used to collect joint status data and end-effector trajectory data of each robotic arm in real time, generate smooth trajectory curves through a fifth-order polynomial interpolation algorithm, and output smooth trajectory data with timestamps.

[0012] The dynamic timing planner is connected to the status monitoring and trajectory smoothing module. It receives the smoothed trajectory data with timestamps, adjusts the timing of the robotic arm's movements in real time based on task progress and space occupancy status, and generates timing optimization instructions.

[0013] An energy consumption balancing optimizer is connected to the dynamic timing planner, receives the timing optimization instructions, calculates the energy consumption distribution of the robotic arm at each trajectory point through a dynamic model, and generates energy consumption balancing trajectory correction parameters.

[0014] The interference detection and correction module is connected to the energy consumption balance optimizer, receives the trajectory correction parameters of the energy consumption balance, detects potential interference between robotic arms in real time through spatial rasterization modeling, generates anti-interference correction vectors in combination with energy consumption optimization requirements, and finally outputs safety control commands for each robotic arm.

[0015] As a further technical solution of the present invention, the state monitoring and trajectory smoothing module includes:

[0016] The multi-source data acquisition unit is used to acquire joint angle, angular velocity and torque data of each robotic arm in real time through encoders and torque sensors;

[0017] The trajectory reconstruction unit is connected to the multi-source data acquisition unit and maps the joint space data into the end-effector Cartesian space trajectory through inverse kinematics calculation.

[0018] A fifth-order polynomial interpolation unit, connected to the trajectory reconstruction unit, performs smooth interpolation on discrete trajectory points to generate smooth trajectory data with timestamps.

[0019] As a further technical solution of the present invention, the trajectory reconstruction unit maps joint space data to an end-effector Cartesian space trajectory through inverse kinematics calculation, specifically including the following steps:

[0020] The link coordinate system of each robotic arm is constructed by using the DH parameter method, and the homogeneous transformation matrix between adjacent links is determined.

[0021] Based on the collected joint angle data, the pose matrix of the end effector relative to the base coordinate system is calculated by chain multiplication.

[0022] Calculate the velocity Jacobian matrix of the robotic arm and establish the velocity mapping relationship between joint space and Cartesian space;

[0023] A closed-loop solution based on geometric analysis is adopted to decompose the desired end-effector Cartesian space trajectory into an angle sequence of each joint. Specifically: for the first three joints of the robotic arm, the wrist center position is determined by geometric projection, and the inverse position solution is solved; for the last three joints of the robotic arm, the end-effector attitude is determined by Euler angle decomposition, and the inverse attitude solution is solved; when there are multiple inverse solutions, the optimal solution is selected based on the principle of minimizing joint movement to ensure a smooth and continuous trajectory.

[0024] The accuracy of the inverse kinematics solution is verified by forward kinematics inverse calculation, ensuring the consistency of the mapping between joint space data and the end-effector Cartesian space trajectory.

[0025] As a further technical solution of the present invention, the dynamic timing planner includes: a task progress monitoring unit, used to evaluate the task completion status and estimated remaining time of each robotic arm in real time;

[0026] The space occupancy analysis unit is connected to the task progress monitoring unit and predicts the space occupancy in future time periods based on the smooth trajectory data with timestamps.

[0027] The timing optimization unit, connected to the space occupancy analysis unit, dynamically adjusts the start time and execution duration of the robotic arm's movements through a conflict detection algorithm, generating timing optimization instructions.

[0028] As a further technical solution of the present invention, the space occupancy analysis unit predicts the space occupancy in the future time period based on smooth trajectory data with timestamps, specifically including the following steps: setting a prediction time window and discretizing the future time period into equally spaced time slices;

[0029] Based on smooth trajectory data with timestamps, the precise poses of each robotic arm link in future time slices are calculated through forward kinematics, and the geometric shape of each link is approximated by a capsule-shaped bounding box.

[0030] The workspace is discretized into a uniform grid in three dimensions. Based on the pose of the robotic arm's surrounding body in each time slice, the set of occupied grids is marked to generate a spatiotemporal occupancy heatmap.

[0031] Considering trajectory tracking error and system delay, a Gaussian process regression model is used to predict the occupancy probability of each grid in different time slices, and a probabilistic spatiotemporal occupancy distribution is established.

[0032] The spatial grid occupancy of multiple robotic arms at the same time was analyzed, and grid overlap areas were identified as potential conflict areas. The time range and severity of the conflict were also marked.

[0033] As a further technical solution of the present invention, the timing optimization unit dynamically adjusts the start time and execution duration of the robotic arm's movements through a conflict detection algorithm to generate timing optimization instructions, specifically including the following steps:

[0034] The system receives potential conflict areas and their corresponding time ranges identified by the space occupancy analysis unit, establishes a conflict time window sequence, and marks the robotic arm identifier and conflict level involved in each conflict.

[0035] Analyze the timing constraints of each robotic arm task to determine the adjustable time range for each action, including the earliest start time, the latest end time, and the maximum extendable duration.

[0036] Dynamic priorities are set based on the urgency of the task and the status of the robotic arm. A time-shifting algorithm is used to adjust the execution time of conflicting actions to a non-overlapping time window, and high-priority tasks maintain their original time sequence.

[0037] For conflicts that cannot be resolved by adjusting the start time, a smooth velocity scaling scheme is generated by adjusting the action execution duration within the allowable range based on the velocity and acceleration constraints of the fifth-order polynomial trajectory model.

[0038] By combining the adjustment results of the start time and execution duration, a timing optimization instruction set is generated, which includes the new start time, execution duration, and speed planning parameters for each robotic arm action.

[0039] As a further technical solution of the present invention, the energy consumption balancing optimizer includes:

[0040] The dynamic modeling unit is used to establish the dynamic equations of the robotic arm and calculate the theoretical energy consumption of the robotic arm at each trajectory point.

[0041] The energy consumption analysis unit is connected to the dynamic modeling unit to analyze the energy consumption distribution characteristics of each robotic arm under the timing optimization command.

[0042] The equilibrium optimization unit, connected to the energy consumption analysis unit, uses a decomposition-based multi-objective evolutionary algorithm to find a balance between energy consumption equilibrium and operational efficiency, and generates trajectory correction parameters for energy consumption equilibrium.

[0043] As a further technical solution of the present invention, the equilibrium optimization unit finds a balance point between energy consumption equilibrium and operational efficiency through a decomposition-based multi-objective evolutionary algorithm, and generates trajectory correction parameters for energy consumption equilibrium, specifically including the following steps:

[0044] Construct a three-objective optimization function that includes total system energy consumption, energy consumption balance among robotic arms, and total task completion time, where energy consumption balance is measured by the standard deviation of the energy consumption values ​​of each robotic arm;

[0045] An initial population is generated in the solution space using a uniform sampling method. Each individual represents a set of possible trajectory correction parameter combinations, including the velocity curves and acceleration constraints of each robotic arm.

[0046] The population individuals are non-dominated and ordered to Pareto levels, and the crowding of individuals within the same level is calculated to maintain solution set diversity.

[0047] A genetic algorithm with an elitist strategy is used to generate offspring populations through selection, crossover, and mutation operations, and non-dominated sorting is performed in combination with the parent populations to retain the optimal Pareto solution set.

[0048] The equilibrium point that best meets the application requirements is selected from the final Pareto front, and the trajectory correction parameters for energy consumption balance are generated based on the individual decoding corresponding to that point.

[0049] As a further technical solution of the present invention, the interference detection and correction module includes:

[0050] An interference prediction unit, connected to the spatial gridding unit, predicts the overlapping area of ​​the robot arm's motion trajectory in the grid space based on the trajectory correction parameters of the energy balance.

[0051] The vector correction unit, connected to the interference prediction unit, calculates the safety avoidance vector between robotic arms using an improved repulsive force field algorithm, generates an anti-interference correction vector by combining energy consumption constraints, and outputs safety control commands for each robotic arm.

[0052] As a further technical solution of the present invention, the vector correction unit calculates the safety avoidance vector between robotic arms through an improved repulsive force field algorithm, generates an anti-interference correction vector by combining energy consumption constraints, and outputs safety control commands for each robotic arm. Specifically, it includes the following steps: establishing an energy consumption-sensitive repulsive force field model and introducing an energy consumption-sensitive factor κ on the basis of the traditional repulsive force field.

[0053] Based on the trajectory correction parameters provided by the energy consumption balance optimizer, the expected energy consumption increment of the robotic arm in different avoidance directions is calculated.

[0054] For each pair of robotic arms that may interfere, calculate the avoidance direction generated by the repulsive force field, and generate a preliminary avoidance vector by combining the kinematic constraints of the robotic arms.

[0055] During the avoidance vector generation process, three objectives are optimized simultaneously: maximizing the safety of avoiding interference, minimizing the additional energy consumption of the avoidance action, and minimizing the impact on trajectory smoothness.

[0056] The finalized anti-interference correction vector is fused with the original trajectory command and converted into safety control commands for each robotic arm joint space through inverse kinematics solution.

[0057] The beneficial effects of this invention are as follows:

[0058] 1. Trajectory smoothing and energy consumption balance optimization: The fifth-order polynomial interpolation algorithm is combined with the decomposition-based multi-objective evolutionary algorithm to achieve energy consumption balance optimization while ensuring the C² continuous trajectory.

[0059] 2. Deep coupling of dynamic timing and energy management: Through real-time interaction between the conflict detection algorithm and the energy analysis model, the energy distribution is optimized synchronously when adjusting the timing of the robotic arm's movements.

[0060] 3. Innovative integration of interference avoidance and energy-sensitive path planning: The improved repulsive force field algorithm introduces an energy-sensitive factor, enabling the system to prioritize the avoidance path with the lowest energy consumption. Attached Figure Description

[0061] Figure 1 This is a block diagram of the multi-robotic arm dynamic timing anti-interference and trajectory smoothing control system proposed in this invention;

[0062] Figure 2 This is an architecture diagram of a dynamic timing planner. Detailed Implementation

[0063] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0064] Please see the appendix Figure 1 -Appendix Figure 2 The multi-robotic arm dynamic timing anti-interference and trajectory smoothing control system includes four core processing modules connected sequentially via a high-speed data bus, specifically:

[0065] S1. Status Monitoring and Trajectory Smoothing Module: This module collects joint status data and end-effector trajectory data of each robotic arm in real time. It generates smooth trajectory curves using a fifth-order polynomial interpolation algorithm and outputs smooth trajectory data with timestamps. Specifically, it includes:

[0066] S11. The multi-source data acquisition unit acquires the joint angle, angular velocity and torque data of each robotic arm in real time through encoders and torque sensors. This unit ensures the accuracy and real-time performance of the data through a high-precision sensor network, providing a reliable data foundation for subsequent trajectory reconstruction and interpolation processing.

[0067] S111. Encoders are installed on each joint of the robotic arm to measure the angular position of the joint. Assuming the first joint... The encoder output of each joint is The joint angle data are then expressed as: ,in: Indicates the first Each joint in time The angle value, in radians; For the first The encoder output values ​​of each joint;

[0068] S112. Angular velocity is obtained by calculating the time derivative of the joint angle data. Assume the angular velocity is... ,but: ,in: Indicates the first Each joint in time Angular velocity, expressed in radians per second;

[0069] S113. A torque sensor is installed at the joint of the robotic arm to measure the torque of the joint. Assume the first... The torque sensor output of each joint is Then the torque data is expressed as: ,in: Indicates the first Each joint in time The torque value, in Newton-meters (Nm). ); For the first The torque sensor output value of each joint.

[0070] S12. The trajectory reconstruction unit maps joint space data to end-effector Cartesian space trajectory through inverse kinematics calculation;

[0071] S121. Construct the link coordinate system of each robotic arm using the DH parameter method. Assume the robotic arm has... There are 1 joint, and the D-H parameters of each joint are as follows: ),in Homogeneous transformation matrix between adjacent links Represented as:

[0072]

[0073] in: The angle is a joint angle, expressed in radians. This is the link torsion angle, expressed in radians. This represents the linkage offset, in meters. The length of the connecting rod is in meters.

[0074] S122. Based on the collected joint angle data The pose matrix of the end effector relative to the base coordinate system is calculated by chain multiplication. : ,in: It is the first Homogeneous transformation matrix of each joint;

[0075] S123. Calculate the speed Jacobian matrix. Establish joint space velocity With the terminal Cartesian space velocity Mapping relationship: ,in: The linear velocity component of the end effector; The angular velocity component of the end effector; The velocity Jacobian matrix has dimensions of . ; For the first Spatial velocity of each joint;

[0076] S124. Using a closed-loop solution method based on geometric analysis, the desired end-point Cartesian space trajectory is decomposed into an angle sequence of each joint;

[0077] For the first three joints of the robotic arm, the center position of the wrist is determined using geometric projection. : ,in: These are the coordinates of the center of the wrist;

[0078] For the last three joints of the robotic arm, the end effector orientation is determined using Euler angle decomposition. : ,in: They are respectively around Euler angles of the axis, Represents the homogeneous transformation matrix The attitude information of the end effector is extracted and represented by three Euler angles;

[0079] When multiple inverse solutions exist, the optimal solution is selected based on the principle of minimizing joint movement. ,in: It is the optimal solution. It is the joint angle at the previous moment;

[0080] S125. Verify the accuracy of the inverse kinematics solution through forward kinematics inverse calculation: ,in: To compare the end pose matrix reconstructed through inverse kinematics. and This ensures consistency between the two. S13. The fifth-order polynomial interpolation unit performs smooth interpolation on discrete trajectory points to generate smooth trajectory data with timestamps. This unit interpolates each trajectory segment using a fifth-order polynomial function to ensure that the trajectory is continuous in position, velocity, and acceleration, thereby achieving smooth motion control. Fifth-order polynomial interpolation can effectively avoid abrupt changes in higher-order derivatives and improve the stability and accuracy of the robotic arm's motion.

[0081] S131. Assume that a time interval is required. Interpolation of the trajectory using a fifth-order polynomial interpolation function. Represented as: ,in: and These are the start and end times of the interpolation interval, respectively. These are the coefficients of the polynomial;

[0082] S132. To ensure that the trajectory is continuous in position, velocity, and acceleration, the following boundary conditions must be met: ,in: For time Interpolation position at that point These are the start and end positions, respectively. For time Interpolation speed at that point These are the starting and ending speeds, respectively. For time Interpolation acceleration at that point These are the initial and final accelerations, respectively.

[0083] S133. Establish a system of linear equations using boundary conditions to solve for the polynomial coefficients. : ;

[0084] S134. Within the time interval Within, for any time Through interpolation function Calculate the corresponding trajectory points: ;

[0085] Simultaneously, calculate velocity and acceleration:

[0086]

[0087] .

[0088] S2. A dynamic timing planner, connected to the status monitoring and trajectory smoothing module, receives the timestamped smoothed trajectory data, adjusts the robotic arm's action timing in real time based on task progress and space occupancy status, and generates timing optimization instructions; specifically including:

[0089] S21. The task progress monitoring unit calculates the task progress and predicts the completion time by monitoring the deviation between the actual movement trajectory of the robotic arm and the preset trajectory;

[0090] Assuming the total task length of the robotic arm is The length of the currently completed task is Then the task progress Represented as: ,in: The range of values ​​is ;

[0091] Estimated remaining time Represented as: ,in: It is the average speed of the robotic arm during the completed task segment.

[0092] S22. The space occupancy analysis unit predicts space occupancy in future time periods based on smoothed trajectory data with timestamps; S221. Set the prediction time window. This represents the time range that needs to be predicted in the future, and it discretizes the future time period into equally spaced time slices. The number of time slices is: ,in: It refers to the number of time slices;

[0093] S222. Based on smooth trajectory data with timestamps, calculate the precise pose of each robotic arm link in future time slices using forward kinematics. For each time slice... Calculate the pose of the robotic arm links : ,in: Is the connecting rod in time slice The homogeneous transformation matrix, Indicates the first Each joint in time The angle values; the geometric shape of each link is approximated by a capsule-shaped enclosure, which consists of two spheres and one cylinder: ,in: Represents the capsule body; Represents a sphere, used to indicate the rounded parts of a robotic arm joint or link; Representing a cylinder, used to indicate the slender main body of a robotic arm link; It is the radius of the capsule body. and These are the two endpoints of the capsule body;

[0094] S223. Workspace The 3D structure is discretized into a uniform grid, with each grid cell having a size of . Total number of grid cells for: ,in: It refers to the dimensions of the workspace;

[0095] For each time slice Based on the position of the robotic arm linkage Mark the set of grid cells that are occupied. First, for each grid g, check if it intersects with the capsule body. If grid g intersects with any part of the capsule body, then the grid is marked as occupied, in time slice. The set of all grids that intersect with the geometry of the link. Recorded: ;

[0096] Generate a spatiotemporal occupancy heatmap Its dimensions are , This indicates the occupancy of each time slice and each grid cell: ;

[0097] S224. Considering trajectory tracking error and system delay, assume the occupancy probability. Following a Gaussian process: ,in: It is a mean function, usually assumed to be a constant; It is a kernel function that represents the correlation between time slices; Represents a Gaussian process;

[0098] Predict the occupancy probability of each grid cell in different time slices using a Gaussian process regression model. : ,in: It is the predicted mean. It is the standard deviation of the prediction. It is a random variable that follows a standard normal distribution.

[0099] S225. Analyze the spatial grid occupancy of multiple robotic arms at the same time, for each time slice. Identify overlapping raster regions As a potential conflict zone, ,

[0100] in: In time slice The sets of overlapping grid cells; and the time range in which the conflicts occurred are indicated. and severity : , ,in: It is the set of time slices in which conflicts occur. It is a time film The severity of the conflict represents the sum of the occupancy probabilities of overlapping graticles.

[0101] S23. The timing optimization unit dynamically adjusts the start time and execution duration of the robotic arm's movements through a conflict detection algorithm, generating timing optimization instructions;

[0102] S231. Receive the potential conflict areas and their corresponding time ranges identified by the space occupancy analysis unit, and establish a conflict time window sequence. , ,in: It is the number of conflict zones. and They are the first The start and end times of each conflict zone; for each conflict zone Mark the robotic arm involved. and conflict level ;

[0103] S232. For each robotic arm Every action Define the following timing constraints:

[0104]

[0105]

[0106] in: It is an action The earliest start time, It is an action The latest start time, It is an action Maximum extendable duration, It is an action The start time, It is an action End time;

[0107] S233. For each robotic arm Every action Set dynamic priority : ,in: It is a function that adjusts the urgency of the task. and robotic arm status Priority is calculated using a time-shifting algorithm for each conflict zone. Adjust the start time of the robotic arm movements involved. and end time This ensures that the action time windows do not overlap.

[0108]

[0109]

[0110] in: It is an action The original execution time, It is the minimum time interval between actions;

[0111] S234. For conflicts that cannot be resolved by adjusting the start time, based on the velocity and acceleration constraints of the fifth-order polynomial trajectory model, assume the action... The trajectory is represented by a fifth-degree polynomial: ,in: These are polynomial coefficients; velocity constraint: , ,in: This is the maximum permissible speed; acceleration constraints: , ,in: It is the maximum permissible acceleration;

[0112] Adjust the execution time of the action within the allowable range. To satisfy the velocity and acceleration constraints, a smooth velocity scaling scheme is generated:

[0113] ,in: It is a variable representing possible candidate values ​​for execution time;

[0114] S235. Generated timing optimization instruction set :

[0115] ,in: It is an action The new start time, It is an action The new execution duration, It is an action The fifth-order polynomial trajectory parameters, It is the total number of robotic arms. It is the total number of actions that each robotic arm needs to perform.

[0116] S3. Energy consumption balancing optimizer, connected to the dynamic timing planner, receives the timing optimization instructions, calculates the energy consumption distribution of the robotic arm at each trajectory point through a dynamic model, and generates energy consumption balancing trajectory correction parameters; specifically including:

[0117] S31. The dynamic modeling unit is used to establish the dynamic equations of the robotic arm and calculate the theoretical energy consumption of the robotic arm at each trajectory point. Through the dynamic model, the energy consumption of the robotic arm under different motion states can be accurately analyzed, providing basic data for energy consumption optimization.

[0118] The dynamic equations of a robotic arm are usually expressed using the Lagrange equations: ,in: It is a joint position vector. It is the joint velocity vector. It is the joint acceleration vector. It is the inertia matrix. It is the matrix of Coriolis force and centrifugal force. It is the gravity vector. It is the joint torque vector;

[0119] Theoretical energy consumption is calculated by integrating the product of joint torque and velocity: ,in: and These are the start and end times of the trajectory. In the time interval Total energy consumption within the unit This represents the transpose of the joint torque vector.

[0120] S32. The energy consumption analysis unit analyzes the energy consumption distribution characteristics of each robotic arm under the timing optimization command. By analyzing the energy consumption distribution, it can identify actions with high energy consumption and provide a basis for subsequent energy consumption optimization.

[0121] For each robotic arm Every action Calculate its energy consumption under timing-optimized instructions. : ,in: and It is an action The start and end times, It is an action The joint torque vector, It is an action The joint velocity vector;

[0122] Analyze energy consumption distribution characteristics, identify high-energy-consuming actions, and calculate statistics such as average energy consumption, maximum energy consumption, and standard deviation of energy consumption for each action:

[0123]

[0124]

[0125]

[0126] in: It is an action Average energy consumption It is an action Maximum energy consumption It is an action The standard deviation of energy consumption.

[0127] S33. The equilibrium optimization unit finds a balance between energy consumption equilibrium and operational efficiency through a decomposition-based multi-objective evolutionary algorithm, and generates trajectory correction parameters for energy consumption equilibrium.

[0128] S331. Total system energy consumption: ,in: It is the first The first robotic arm Energy consumption of each action;

[0129] The energy consumption balance among robotic arms is measured by the standard deviation of the energy consumption values ​​of each robotic arm. ,in: It is the first Total energy consumption of the robotic arm , This is the average energy consumption of all robotic arms; It is the average energy consumption of all robotic arms. ;

[0130] Total task completion time: ,in: It is the first The first robotic arm The end time of each action.

[0131] Construct a three-objective optimization function: .

[0132] S332. In the solution space Initial population generated by uniform sampling : ,in: It is the population size. It is the first Each individual represents a set of trajectory correction parameter combinations; each individual Includes the velocity curves and acceleration constraints of each robotic arm: ,in: It is the first The first robotic arm The speed curve of each action, It is the first The first robotic arm Acceleration constraints for each action.

[0133] S333. On the population Perform non-dominated sorting and assign Pareto levels. :

[0134] Includes all individuals that are not controlled by other individuals. Included in removal Individuals that are not controlled by other individuals are assigned to a Pareto level, and so on, until all individuals are assigned to a certain Pareto level.

[0135] At the same level Internal calculation of individual crowding : ,in: Individual In the Proximity distances on each target It is the first The range of values ​​for each target.

[0136] S334. Using a genetic algorithm with an elitist strategy, first start from the current population... Select an individual to generate a progeny population Then, the offspring population Perform crossover operations to generate new offspring populations Finally, for the new offspring population Mutation operations are performed to generate the final offspring population. ;

[0137] parental population and offspring population Combine, perform non-dominated sorting, and retain the optimal Pareto solution set. .

[0138] S335. From the final Pareto frontier Select the balance point that best meets the application requirements. : ,in: These are weighting coefficients, adjusted according to specific application requirements; based on the optimal solution. Decode and generate trajectory correction parameters for energy balance: ,in: For the first The first robotic arm The optimal speed curve for each action. For the first The first robotic arm The optimal acceleration constraint for each action, and Decode is the decoding function that converts the output of the optimization algorithm into specific trajectory correction parameters.

[0139] S4. Interference detection and correction module, connected to the energy consumption balance optimizer, receives trajectory correction parameters from the energy consumption balance, detects potential interference between robotic arms in real time through spatial rasterization modeling, generates anti-interference correction vectors based on energy consumption optimization requirements, and finally outputs safety control commands for each robotic arm; specifically including:

[0140] S41. The interference prediction unit predicts the overlapping area of ​​the robot arm's motion trajectory in the grid space based on the trajectory correction parameters of energy consumption balance. By analyzing the motion trajectory of each robot arm, it identifies potential interference areas and provides data support for subsequent interference correction.

[0141] For each time point Predicting robotic arms and Overlapping area of ​​motion trajectories : ,in: and These are robotic arms and In time The set of occupied grid cells It is two robotic arms in time An overlapping set of grid cells;

[0142] Define potential interference regions The union of the overlapping regions at all time points: ,in: Includes all possible interference grids;

[0143] Define the time range of interference The set of time points where overlapping regions exist: ,in: This represents all points in time where potential interference exists.

[0144] S42. The vector correction unit calculates the safety avoidance vector between robotic arms through an improved repulsive force field algorithm, generates an anti-interference correction vector by combining energy consumption constraints, and outputs safety control commands for each robotic arm.

[0145] S421. Establish an energy-sensitive repulsive force field model, introducing an energy-sensitive factor κ based on the traditional repulsive force field, and expressing the magnitude of the repulsive force as: when d≤ When d> , Where: d is the actual distance between the robotic arms; To influence the distance threshold, ∇d represents the distance gradient; As an energy-sensitive factor, Set as a function negatively correlated with the expected energy consumption increment: ,in: It is the baseline sensitivity coefficient. It's about adjusting parameters. This is the expected increase in energy consumption. It is an exponential function;

[0146] S422. Based on the trajectory correction parameters provided by the energy consumption balance optimizer, for each avoidance direction... Calculate the expected energy consumption increment : ,in: In the direction of avoidance The joint torque below, In the direction of avoidance The joint velocity of the lower joint, and These are the joint torques and velocities under the original trajectory. for The transpose of the matrix;

[0147] S423. For each robotic arm pair that may interfere... Calculate the avoidance direction generated by the repulsive force field. : ,in: It is in time At that time, the robotic arm and The repulsive force between them;

[0148] Based on the kinematic constraints of the robotic arm, a preliminary avoidance vector is generated. : ,in: It is a robotic arm The set of kinematic constraints; It is a projection operation that ensures the avoidance vector conforms to kinematic constraints;

[0149] S424. Simultaneously optimize three objectives during the avoidance vector generation process: maximizing safety, i.e., maximizing the magnitude of the avoidance vector: Minimizing additional energy consumption means minimizing the additional energy consumption of the avoidance maneuver. Minimizing the impact of trajectory smoothness means minimizing the influence of trajectory smoothness: ,in: It is the velocity vector under the original trajectory;

[0150] S425. The final determined anti-interference correction vector With the original trajectory instructions Fusion: By solving inverse kinematics, the fused velocity vector is obtained. Velocity commands converted to joint space : ,in: It is a robotic arm Jacobian matrix, It is the pseudo-inverse of the Jacobian matrix.

[0151] Example 1

[0152] 1. Test Environment and Conditions: The hardware environment consisted of 8 six-DOF industrial robotic arms with a workspace overlap of 85%; the sensor system comprised high-precision encoders (resolution 0.001°) and torque sensors (accuracy ±0.1 N·m); the control cycle was a 1 ms real-time control cycle; the test scenario was a simulated automotive engine assembly line environment. The complex assembly task involved multi-robotic arm collaborative operation with 32 assembly steps; the high-density path involved an average spacing of ≤150 mm between the robotic arm end-effector trajectories; dynamic disturbances simulated sudden task insertions and priority adjustments in actual production.

[0153] 2. Implementation steps and testing methods

[0154] 2.1 Implementation of the System of the Invention

[0155] Step 1: System initialization: Load the DH parameters of each robotic arm and establish the kinematic model; configure the fifth-order polynomial interpolation parameters: maximum acceleration 3m / s², maximum jerk 15m / s³; set the energy consumption balance weights: energy efficiency 0.4, balance degree 0.3, and operation efficiency 0.3.

[0156] Step 2: Trajectory Planning and Optimization: The state monitoring module collects the initial joint state and generates a C² continuous trajectory; the dynamic timing planner adjusts the action timing based on space occupancy prediction; the energy consumption balance optimizer calculates trajectory correction parameters through a decomposition-based multi-objective evolutionary algorithm.

[0157] Step 3: Real-time control and monitoring: The interference detection module updates the safety distance field at a 10ms cycle; the vector correction unit generates anti-interference commands based on the energy-sensitive repulsive force field; and records trajectory tracking errors, energy consumption data, and interference early warning information in real time.

[0158] 2.2 Comparison of system configurations: Traditional system A is based on cubic spline interpolation + fixed-time programming; traditional system B is a real-time obstacle avoidance system based on artificial potential field method; the test duration is 24 hours of continuous operation for each system, and the test is repeated 3 times.

[0159] 3. Performance Comparison Results: The performance comparison between the present invention and the prior art is shown in Table 1 below:

[0160] Table 1: Performance Comparison Table

[0161]

[0162] The system of this invention achieves continuous smoothing of position, velocity, and acceleration at three levels through a fifth-order polynomial interpolation algorithm: the peak acceleration is reduced from 15.2 m / s² in the traditional system to 6.8 m / s², a reduction of 55.3%; the measured vibration of the robotic arm base is reduced from 0.12 mm to 0.05 mm, a reduction of 58.3%; and the motion impact noise is reduced from 85 dB to 72 dB, improving the working environment.

[0163] The application of the energy balance optimizer brings significant benefits: the standard deviation of the running time of each robotic arm is reduced from 18.7% in the traditional system to 7.2%; the maximum temperature rise is reduced from 45°C to 28°C, extending the equipment life; and the peak power demand is reduced by 33.3%, reducing the impact on the power grid.

[0164] The improved repulsive force field algorithm, combined with energy consumption constraints, enables intelligent avoidance: the avoidance path is optimized, the average avoidance path length is shortened by 42%, and the time is reduced by 55%; the production line downtime caused by interference is reduced from 3.2% in the traditional system to 0.3%; safety is improved, achieving a 24-hour zero-collision operation record.

[0165] 4. Validation in special scenarios

[0166] High dynamic environment adaptability test: Simulating a sudden task insertion scenario, the system of this invention exhibits superior performance: replanning response time, only 125ms from task change to new trajectory generation; smooth transition, trajectory jump during task switching is <0.5mm; system stability, still maintaining stable operation when the task load changes abruptly by 30%.

[0167] Long-term operational reliability verification: The results of a continuous 720-hour endurance test show that the decay rate of key performance indicators is <2%, which is significantly better than the 8-12% of traditional systems; the preventive maintenance cycle is extended from 500 hours to 1200 hours of traditional systems; the system hardware failure rate is reduced by 65%, and the software anomaly is reduced by 82%.

[0168] 5. Industrial Application Value Assessment: Based on the test results, the system of this invention has significant value in industrial applications: it is expected to improve the overall efficiency of the assembly line by more than 33%; save approximately 9,800 kWh of electricity per year and reduce carbon emissions by 7.8 tons; extend the lifespan of key mechanical components by more than 40%; and essentially eliminate production accidents caused by robotic arm interference.

[0169] As can be seen from the above description, the embodiments of the present invention achieve the following technical effects:

[0170] Breakthrough in the Co-optimization of Trajectory Smoothness and Energy Consumption Balance: This invention deeply couples a fifth-order polynomial interpolation algorithm with an energy consumption balance optimizer, resolving the traditional contradiction between trajectory smoothness and energy consumption optimization. By employing a decomposition-based multi-objective evolutionary algorithm to find the optimal balance between energy consumption balance and operational efficiency, the system achieves triple synergistic optimization: While ensuring C² continuous trajectory, the range of joint torque fluctuations in the robotic arm is reduced by 50%, significantly reducing motion impact; the multi-objective optimization model simultaneously optimizes the system's total energy consumption, energy consumption balance, and task completion time, improving the energy consumption balance of the 8-robotic arm system by 42% and reducing total energy consumption by 28%; the introduced energy consumption sensitive factor mechanism allows the system to automatically prioritize the path with the smallest energy consumption increment during trajectory correction, avoiding the sharp energy consumption increase phenomenon common in traditional anti-interference strategies. This synergistic optimization results in an overall energy efficiency improvement of 33% compared to traditional methods and a 2.5-fold increase in equipment maintenance intervals during a 48-hour continuous operation test.

[0171] A deep integration of dynamic time-series planning and energy management innovation: The system achieves unified optimal allocation of time and energy resources through real-time interaction between a dynamic time-series planner and an energy balance optimizer. Based on a time window reallocation mechanism using a conflict detection algorithm, combined with adaptive execution duration adjustment technology, the system can dynamically adjust the timing of robotic arm movements according to the real-time task status, reducing the spatial conflict rate in high-density layouts from 18.7% to 2.3%. Energy consumption constraints are embedded in the time-series decision-making process. Through a dynamic model established by the energy consumption analysis unit, the system predicts the energy consumption distribution under different time-series schemes, ensuring that time-series optimization considers not only time efficiency but also energy economy.

[0172] An innovative integration of interference avoidance and energy-sensitive path planning: This invention integrates energy consumption constraints directly into the core of anti-interference decision-making through an improved repulsive force field algorithm, innovatively solving the problem of balancing safety and energy efficiency in traditional methods. Building upon traditional distance-sensitive methods, this algorithm introduces an energy-sensitive factor, making the repulsive force field strength negatively exponentially related to the expected energy consumption increment. When potential interference is detected, the system automatically prioritizes the lowest-energy-consumption avoidance path. Furthermore, by combining spatial rasterization modeling with probabilistic spatiotemporal occupancy prediction, the system can predict conflict areas within future time windows, providing sufficient response time for energy-optimal avoidance decisions and ensuring the timeliness and economy of anti-interference correction. This integrated innovation enables the system to simultaneously achieve zero-collision safety records and significant energy savings in high-density dynamic environments.

[0173] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of the invention is limited to these examples; within the framework of the invention, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of the different aspects of the invention as described above, which are not provided in detail for the sake of brevity.

[0174] This invention is intended to cover all such substitutions, modifications, and variations that fall within the broad scope of this specification. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A multi-robotic arm dynamic timing anti-interference and trajectory smoothing control system, characterized in that, It includes four core processing modules connected sequentially via a high-speed data bus, specifically: The status monitoring and trajectory smoothing module is used to collect joint status data and end-effector trajectory data of each robotic arm in real time, generate smooth trajectory curves through a fifth-order polynomial interpolation algorithm, and output smooth trajectory data with timestamps. The dynamic timing planner is connected to the status monitoring and trajectory smoothing module. It receives the smoothed trajectory data with timestamps, adjusts the timing of the robotic arm's movements in real time based on task progress and space occupancy status, and generates timing optimization instructions. An energy consumption balancing optimizer is connected to the dynamic timing planner, receives the timing optimization instructions, calculates the energy consumption distribution of the robotic arm at each trajectory point through a dynamic model, and generates energy consumption balancing trajectory correction parameters. The interference detection and correction module is connected to the energy consumption balance optimizer, receives the trajectory correction parameters of the energy consumption balance, detects potential interference between robotic arms in real time through spatial rasterization modeling, generates anti-interference correction vectors in combination with energy consumption optimization requirements, and finally outputs safety control commands for each robotic arm.

2. The multi-robotic arm dynamic timing anti-interference and trajectory smoothing control system according to claim 1, characterized in that, The status monitoring and trajectory smoothing module includes: The multi-source data acquisition unit is used to acquire joint angle, angular velocity and torque data of each robotic arm in real time through encoders and torque sensors; The trajectory reconstruction unit is connected to the multi-source data acquisition unit and maps the joint space data into the end-effector Cartesian space trajectory through inverse kinematics calculation. A fifth-order polynomial interpolation unit, connected to the trajectory reconstruction unit, performs smooth interpolation on discrete trajectory points to generate smooth trajectory data with timestamps.

3. The multi-robotic arm dynamic timing anti-interference and trajectory smoothing control system according to claim 2, characterized in that, The trajectory reconstruction unit maps joint space data to an end-effector Cartesian space trajectory through inverse kinematics calculations, specifically including the following steps: The link coordinate system of each robotic arm is constructed by using the DH parameter method, and the homogeneous transformation matrix between adjacent links is determined. Based on the collected joint angle data, the pose matrix of the end effector relative to the base coordinate system is calculated by chain multiplication. Calculate the velocity Jacobian matrix of the robotic arm and establish the velocity mapping relationship between joint space and Cartesian space; A closed-loop solution based on geometric analysis is used to decompose the desired terminal Cartesian space trajectory into an angle sequence of each joint. The accuracy of the inverse kinematics solution is verified by forward kinematics inverse calculation, ensuring the consistency of the mapping between joint space data and the end-effector Cartesian space trajectory.

4. The multi-robotic arm dynamic timing anti-interference and trajectory smoothing control system according to claim 1, characterized in that, The dynamic timing planner includes a task progress monitoring unit, which is used to evaluate the task completion status and estimated remaining time of each robotic arm in real time. The space occupancy analysis unit is connected to the task progress monitoring unit and predicts the space occupancy in future time periods based on the smooth trajectory data with timestamps. The timing optimization unit, connected to the space occupancy analysis unit, dynamically adjusts the start time and execution duration of the robotic arm's movements through a conflict detection algorithm, generating timing optimization instructions.

5. The multi-robotic arm dynamic timing anti-interference and trajectory smoothing control system according to claim 4, characterized in that, The space occupancy analysis unit predicts the space occupancy in future time periods based on smooth trajectory data with timestamps, specifically including the following steps: setting a prediction time window and discretizing the future time period into equally spaced time slices; Based on smooth trajectory data with timestamps, the precise poses of each robotic arm link in future time slices are calculated through forward kinematics, and the geometric shape of each link is approximated by a capsule-shaped bounding box. The workspace is discretized into a uniform grid in three dimensions. Based on the pose of the robotic arm's surrounding body in each time slice, the set of occupied grids is marked to generate a spatiotemporal occupancy heatmap. Considering trajectory tracking error and system delay, a Gaussian process regression model is used to predict the occupancy probability of each grid in different time slices, and a probabilistic spatiotemporal occupancy distribution is established. The spatial grid occupancy of multiple robotic arms at the same time is analyzed, and grid overlap areas are identified as potential conflict areas. The time range and severity of the conflict are also marked.

6. The multi-robotic arm dynamic timing anti-interference and trajectory smoothing control system according to claim 5, characterized in that, The timing optimization unit dynamically adjusts the start time and execution duration of the robotic arm's movements using a conflict detection algorithm to generate timing optimization instructions, specifically including the following steps: The system receives potential conflict areas and their corresponding time ranges identified by the space occupancy analysis unit, establishes a conflict time window sequence, and marks the robotic arm identifier and conflict level involved in each conflict. Analyze the timing constraints of each robotic arm task to determine the adjustable time range for each action, including the earliest start time, the latest end time, and the maximum extendable duration. Dynamic priorities are set based on the urgency of the task and the status of the robotic arm. A time-shifting algorithm is used to adjust the execution time of conflicting actions to a non-overlapping time window, and high-priority tasks maintain their original time sequence. For conflicts that cannot be resolved by adjusting the start time, a smooth velocity scaling scheme is generated by adjusting the action execution duration within the allowable range based on the velocity and acceleration constraints of the fifth-order polynomial trajectory model. By combining the adjustment results of the start time and execution duration, a timing optimization instruction set is generated, which includes the new start time, execution duration, and speed planning parameters for each robotic arm action.

7. The multi-robotic arm dynamic timing anti-interference and trajectory smoothing control system according to claim 1, characterized in that, The energy consumption balancing optimizer includes: The dynamic modeling unit is used to establish the dynamic equations of the robotic arm and calculate the theoretical energy consumption of the robotic arm at each trajectory point. The energy consumption analysis unit is connected to the dynamic modeling unit to analyze the energy consumption distribution characteristics of each robotic arm under the timing optimization command. The equilibrium optimization unit, connected to the energy consumption analysis unit, uses a decomposition-based multi-objective evolutionary algorithm to find a balance between energy consumption equilibrium and operational efficiency, and generates trajectory correction parameters for energy consumption equilibrium.

8. The multi-robotic arm dynamic timing anti-interference and trajectory smoothing control system according to claim 7, characterized in that, The equilibrium optimization unit uses a decomposition-based multi-objective evolutionary algorithm to find a balance between energy consumption equilibrium and operational efficiency, generating trajectory correction parameters for energy consumption equilibrium. Specifically, this includes the following steps: Construct a three-objective optimization function that includes total system energy consumption, energy consumption balance among robotic arms, and total task completion time, where energy consumption balance is measured by the standard deviation of the energy consumption values ​​of each robotic arm; An initial population is generated in the solution space using a uniform sampling method. Each individual represents a set of possible trajectory correction parameter combinations, including the velocity curves and acceleration constraints of each robotic arm. The population individuals are non-dominated and ordered to Pareto levels, and the crowding of individuals within the same level is calculated to maintain solution set diversity. A genetic algorithm with an elitist strategy is used to generate offspring populations through selection, crossover, and mutation operations, and non-dominated sorting is performed in combination with the parent populations to retain the optimal Pareto solution set. The equilibrium point that best meets the application requirements is selected from the final Pareto front, and the trajectory correction parameters for energy consumption balance are generated based on the individual decoding corresponding to that point.

9. The multi-robotic arm dynamic timing anti-interference and trajectory smoothing control system according to claim 1, characterized in that, The interference detection and correction module includes: An interference prediction unit, connected to the spatial gridding unit, predicts the overlapping area of ​​the robot arm's motion trajectory in the grid space based on the trajectory correction parameters of the energy balance. The vector correction unit, connected to the interference prediction unit, calculates the safety avoidance vector between robotic arms using an improved repulsive force field algorithm, generates an anti-interference correction vector by combining energy consumption constraints, and outputs safety control commands for each robotic arm.

10. The multi-robotic arm dynamic timing anti-interference and trajectory smoothing control system according to claim 9, characterized in that, The vector correction unit calculates the safety avoidance vector between robotic arms through an improved repulsive force field algorithm, generates an anti-interference correction vector by combining energy consumption constraints, and outputs safety control commands for each robotic arm. Specifically, it includes the following steps: establishing an energy consumption-sensitive repulsive force field model and introducing an energy consumption-sensitive factor κ on the basis of the traditional repulsive force field. Based on the trajectory correction parameters provided by the energy consumption balance optimizer, the expected energy consumption increment of the robotic arm in different avoidance directions is calculated. For each pair of robotic arms that may interfere, calculate the avoidance direction generated by the repulsive force field, and generate a preliminary avoidance vector by combining the kinematic constraints of the robotic arms. During the avoidance vector generation process, three objectives are optimized simultaneously: maximizing the safety of avoiding interference, minimizing the additional energy consumption of the avoidance action, and minimizing the impact on trajectory smoothness. The finalized anti-interference correction vector is fused with the original trajectory command and converted into safety control commands for each robotic arm joint space through inverse kinematics solution.

Images