A path planning method for a robotic arm platform, a robotic arm platform, and a storage medium.

Abstract

This invention discloses a path planning method for a robotic arm platform, a robotic arm platform, and a storage medium, relating to the field of robotic arm path planning technology, and is invented to improve path planning accuracy. The method involves acquiring the current pose of the robotic arm platform, current environmental information, and the target arrival point. An initial obstacle avoidance path is generated using a gradient descent method with the adversarial potential field as parameters, taking the current environmental information and the target arrival point as parameters. This initial obstacle avoidance path is then smoothed using a filtering algorithm to obtain a smoothed obstacle avoidance path. The dynamic model of the robotic arm platform is used as a constraint, and an iterative optimization of the smoothed obstacle avoidance path is performed using a target value function with tracking error penalty, smoothness penalty, and energy consumption penalty as parameters, resulting in the planned path for the robotic arm platform after the target value function converges. This invention is used for robotic arm path planning.

Description

Technical Field

[0001] This invention relates to the field of robotic arm path planning technology, and in particular to a robotic arm platform path planning method, a robotic arm platform, and a storage medium. Background Technology

[0002] Vehicle-mounted mobile robotic arms are core automated equipment in coal mine roadway cleaning operations. Past research has proposed many methods for robotic arm path planning, but all have various shortcomings. Although existing technologies have made some progress in robotic arm path planning, the following significant drawbacks remain: First, most methods fail to fully consider the dynamic combined effect between the vehicle-mounted mobile platform and the robotic arm. Specifically, the actual motion accuracy of the robotic arm is influenced by both changes in platform pose and joint motion. Ignoring this characteristic can easily lead to significant deviations in the planned path during actual execution, resulting in incomplete cleaning coverage, severe trajectory jitter, and other problems, thus reducing the actual operational quality of the system.

[0003] Secondly, traditional path planning methods, such as the artificial potential field method, exhibit significant limitations when facing unstructured tunnel environments. The artificial potential field method is prone to getting trapped in local minima, leading to path planning failure, while graph search-based methods struggle to handle dynamic obstacles, and their computational efficiency drops sharply with increasing environmental complexity. Furthermore, most methods lack a comprehensive consideration of the robotic arm's dynamic constraints and motion stability, resulting in feasible paths with poor control accuracy.

[0004] In addition, existing optimization methods often focus on a single performance index and lack a systematic framework for the coordinated optimization of multiple objectives such as trajectory accuracy, motion stability and energy efficiency, which can easily lead to performance degradation of the system under complex working conditions. Summary of the Invention

[0005] In view of the above problems, embodiments of the present invention provide a path planning method for a robotic arm platform, a robotic arm platform, and a computer-readable storage medium, which can improve the accuracy of robotic arm path planning.

[0006] In a first aspect, embodiments of the present invention provide a path planning method for a robotic arm platform, comprising: The current pose, current environment information, and target arrival point of the robotic arm platform are obtained. The initial obstacle avoidance path is generated by using the gradient descent method with the current environment information and the target arrival point as parameters to generate the counteracting resultant potential field. The initial obstacle avoidance path is smoothed using a filtering algorithm to obtain a smoothed obstacle avoidance path. The dynamic model of the robotic arm platform is used as a constraint, and the objective value function with tracking error penalty, smoothness penalty and energy consumption penalty as parameters is iterated to optimize the smoothed obstacle avoidance path, so as to obtain the robotic arm platform planning path corresponding to the convergence of the objective value function. The planned path of the robotic arm platform is converted into a continuous motion trajectory, and the robotic arm is driven to perform continuous motion by the robotic arm platform control module according to the continuous motion trajectory.

[0007] In one implementation, the counteracting resultant potential field, using the current environmental information and the target arrival point as parameters, includes: Using the target arrival point as a reference, the attraction potential energy is used to solve the quadratic function composed of the starting point of the robotic arm and the target arrival point of the robotic arm, so as to obtain the attraction potential field of the target arrival point of the robotic arm. The repulsive potential energy of the obstacle information on the robot arm in the current environment is taken as a reference. The inverse proportional function established with the shortest distance from the robot arm to the obstacle and the range of the repulsive force of the obstacle on the robot arm is solved to obtain the repulsive potential field of the obstacle. The attractive potential field at the target arrival point of the robotic arm and the repulsive potential field of the obstacle are combined to obtain the counteracting resultant potential field.

[0008] In one implementation, generating an initial obstacle avoidance path using a gradient descent method from the adversarial potential field, with the current environmental information and the target arrival point as parameters, includes: The current pose of the robotic arm platform is used as the starting node, and the motion direction corresponding to the starting node is calculated through the opposing resultant potential field. The next pose of the robotic arm platform is generated by moving along the direction corresponding to the starting node and increasing by a preset step size. Determine whether the next pose of the robotic arm platform will collide with an obstacle; If there is no collision, the next pose of the robotic arm platform is added to the path node to obtain the next moment node, and a termination judgment is made.

[0009] In one implementation, generating an initial obstacle avoidance path using a gradient descent method from the adversarial potential field, with the current environmental information and the target arrival point as parameters, includes: The current pose of the robotic arm platform is used as the starting node, and the motion direction corresponding to the starting node is calculated through the opposing resultant potential field. The next pose of the robotic arm platform is generated by moving along the direction corresponding to the starting node and increasing by a preset step size. Determine whether the next pose of the robotic arm platform will collide with an obstacle; If a collision occurs, the preset step size is reduced, the next pose of the robotic arm platform is regenerated, and it is re-determined whether the next pose of the robotic arm platform will collide with an obstacle.

[0010] In one implementation, defining the dynamic model of the robotic arm platform as a constraint includes: In the dynamic model of the robotic arm platform, the joint torque of the robotic arm in the robotic arm platform is constrained. The joints and angular velocities of the robotic arm in the robotic arm platform do not exceed physical limit constraints; In the robotic arm platform, the distance between the robotic arm and the obstacle is greater than the safe distance threshold constraint.

[0011] In one embodiment, the step of converting the planned path of the robotic arm platform into a continuous motion trajectory, and driving the robotic arm to perform continuous motion by the robotic arm platform control module according to the continuous motion trajectory, includes: The discrete path node sequence in the planned path of the robotic arm platform is converted into a continuous motion trajectory, and corresponding real-time robotic arm joint speed commands and robotic arm acceleration commands are generated. Based on the joint speed command and acceleration command of the robotic arm, the robotic arm platform control module drives the robotic arm to perform continuous movement.

[0012] In a second aspect, embodiments of the present invention provide a robotic arm platform, comprising: The initial planning module is used to obtain the current pose of the robotic arm platform, the current environmental information, and the target arrival point. It generates an initial obstacle avoidance path using the gradient descent method based on the adversarial potential field with the current environmental information and the target arrival point as parameters. The path smoothing module is used to smooth the initial obstacle avoidance path using a filtering algorithm to obtain a smoothed obstacle avoidance path. The path optimization module is used to take the dynamic model of the robotic arm platform as a constraint and iterate the objective value function with tracking error penalty, smoothness penalty and energy consumption penalty as parameters to optimize the smoothed obstacle avoidance path, so as to obtain the robotic arm platform planning path corresponding to the convergence of the objective value function. The execution module converts the planned path of the robotic arm platform into a continuous motion trajectory, and the robotic arm platform control module drives the robotic arm to perform continuous motion according to the continuous motion trajectory.

[0013] Thirdly, a computer-readable storage medium stores one or more programs that can be executed by one or more processors to implement the robotic arm platform path planning method described in any of the foregoing embodiments.

[0014] This invention provides a path planning method for a robotic arm platform, a robotic arm platform, and a computer-readable storage medium. It achieves real-time obstacle avoidance and path generation through an artificial potential field, and combines this with a multi-objective optimization control algorithm to effectively balance the relationship between trajectory accuracy, motion stability, and energy efficiency. This technical solution fully considers the dynamic constraints of the robotic arm and environmental uncertainties, significantly improving the system's operational performance and control quality in confined roadway environments, and providing technical support for achieving efficient and reliable automated cleaning of coal mine roadways. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 A flowchart illustrating a path planning method for a robotic arm platform provided as an embodiment of the present invention; Figure 2 A schematic diagram showing the cross-sectional comparison of the tunnel wall and the trajectory plane before and after smoothing treatment, provided for embodiments of the present invention; Figure 3 A schematic diagram of the modules of a robotic arm platform provided for an embodiment of the present invention; Figure 4 A comparison diagram of the joint angle curve before and after optimization in a rectangular tunnel environment is provided for embodiments of the present invention; Figure 5 A comparison graph of joint angular velocity changes before and after optimization, provided for embodiments of the present invention; Figure 6 A comparison chart of joint angular acceleration changes before and after optimization in a rectangular tunnel environment, provided for embodiments of the present invention; Figure 7 The joint torque required for each joint to perform its task under no-load conditions is provided in the embodiments of the present invention. Detailed Implementation

[0017] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0018] It should be understood that the described embodiments are merely some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0019] Example 1 In a first aspect, embodiments of the present invention provide a path planning method for a robotic arm platform, which achieves real-time obstacle avoidance and path generation through an artificial potential field, and then combines a multi-objective optimization control algorithm to effectively balance the relationship between trajectory accuracy, motion stability and energy efficiency; significantly improving the system's operational performance and control quality in confined roadway environments, and providing technical support for achieving efficient and reliable automated cleaning of coal mine roadways.

[0020] like Figure 1 As shown, an embodiment of the present invention provides a path planning method for a robotic arm platform, comprising: Step S10: Obtain the current pose of the robotic arm platform, the current environment information, and the target arrival point. Generate an initial obstacle avoidance path using the gradient descent method with the current environment information and the target arrival point as parameters for the adversarial resultant potential field. Step S20: Smooth the initial obstacle avoidance path using a filtering algorithm to obtain a smoothed obstacle avoidance path; Step S30: Using the dynamic model of the robotic arm platform as a constraint, and using the target value function with tracking error penalty term, smoothness penalty term and energy consumption penalty term as parameters, iterate to optimize the smoothed obstacle avoidance path, and obtain the robotic arm platform planning path corresponding to the convergence of the target value function. Step S40: Convert the planned path of the robotic arm platform into a continuous motion trajectory, and drive the robotic arm to perform continuous motion according to the continuous motion trajectory by the robotic arm platform control module.

[0021] In actual coal mine roadways, 3D modeling of the coal mine roadway cleaning environment must comprehensively consider roadway structure, wall characteristics, obstacle distribution, and environmental factors to ensure the model's realism and applicability. Common coal mine roadway environments include semi-circular and rectangular roadways. Throughout the modeling process, the roadway wall morphology and potential obstacles (pipelines, cables, supports, etc.) must also be considered to accurately reflect the working environment of the coal mine roadway. Based on the above actual environmental conditions, the robotic arm platform adopts a path planning and control strategy to achieve high-precision smooth motion. Firstly, during path planning, a collaborative processing architecture is used, employing an artificial potential field method for global planning. This involves constructing attractive and repulsive forces to generate an initial path that simultaneously optimizes travel distance and obstacle avoidance. Specifically, when using the artificial potential field method for global planning, the current pose of the robotic arm platform, current environmental information, and target arrival point must first be obtained. Step S10, the counteracting resultant potential field using the current environmental information and the target arrival point as parameters, includes: The counteracting resultant potential field, which uses the current environmental information and the target arrival point as parameters, includes: Using the target arrival point as a reference, the attraction potential energy is used to solve the quadratic function composed of the starting point of the robotic arm and the target arrival point of the robotic arm, so as to obtain the attraction potential field of the target arrival point of the robotic arm. The repulsive potential energy of the obstacle information on the robot arm in the current environment is taken as a reference. The inverse proportional function established with the shortest distance from the robot arm to the obstacle and the range of the repulsive force of the obstacle on the robot arm is solved to obtain the repulsive potential field of the obstacle. The attractive potential field at the target arrival point of the robotic arm and the repulsive potential field of the obstacle are combined to obtain the counteracting resultant potential field.

[0022] Specifically, the three steps for global planning of an artificial potential field are as follows: Constructing an attractive potential field: Attractive potential field As the robotic arm on the platform approaches the target, the size decreases, typically represented by a quadratic function: in, This is the current position of the robotic arm's end effector. The coordinates of the target point, The attraction gain coefficient is the target point attraction. The formula for calculation is: Constructing a repulsive potential field: the repulsive potential energy exerted by an obstacle on the robotic arm. The repulsive field model is usually established using an inverse proportional function, expressed as: in, The shortest distance from the end effector of the robotic arm to the obstacle. For the repulsive force operation range, Let be the repulsive force coefficient; this is the repulsive force of the obstacle. The calculation formula is: Constructing a potential field of opposing forces For the synthesis of attractive and repulsive forces: By designing various typical obstacle distribution scenarios in coal mine roadways, the obstacle avoidance algorithm based on artificial potential field is simulated and verified to evaluate the algorithm's adaptability and stability in complex environments.

[0023] In step S10, generating an initial obstacle avoidance path using the gradient descent method from the adversarial potential field with the current environmental information and the target arrival point as parameters includes: The current pose of the robotic arm platform is used as the starting node, and the motion direction corresponding to the starting node is calculated through the opposing resultant potential field. The next pose of the robotic arm platform is generated by moving along the direction corresponding to the starting node and increasing by a preset step size. Determine whether the next pose of the robotic arm platform will collide with an obstacle; If there is no collision, the next pose of the robotic arm platform is added to the path node to obtain the next moment node, and a termination judgment is made.

[0024] Specifically, a 3D model of the coal mine roadway environment is created, and an artificial potential field algorithm is used to construct the environmental potential field model. By designing a reasonable gravitational field function, the robotic arm is guided to move towards the target point, while a repulsive field function is used to achieve dynamic obstacle avoidance, generating an initial collision-free path that meets real-time obstacle avoidance requirements.

[0025] In another implementation, step S10, which involves generating an initial obstacle avoidance path using the gradient descent method from the adversarial potential field with the current environmental information and the target arrival point as parameters, includes: The current pose of the robotic arm platform is used as the starting node, and the motion direction corresponding to the starting node is calculated through the opposing resultant potential field. The next pose of the robotic arm platform is generated by moving along the direction corresponding to the starting node and increasing by a preset step size. Determine whether the next pose of the robotic arm platform will collide with an obstacle; If a collision occurs, the preset step size is reduced, the next pose of the robotic arm platform is regenerated, and it is re-determined whether the next pose of the robotic arm platform will collide with an obstacle.

[0026] Specifically, before generating an initial collision-free path, it's necessary to determine whether an obstacle has collided. If an obstacle has collided, the step size is reduced, the pose is regenerated, and the obstacle collision check is repeated; that is, it's determined whether the distance between the robotic arm and the obstacle is greater than a safety threshold. ;in, The coordinates of the robotic arm's end effector position. The three-dimensional spatial coordinates of the obstacle The minimum safe distance between the end effector of the robotic arm and the obstacle is defined as follows: the robotic arm is within the safe threshold.

[0027] Furthermore, it is determined whether the difference between the next time node and the target arrival point is within a preset threshold. If so, a set of initial path node sequences and a set of joint angle sequences of the robotic arm platform are obtained, and the generation of a new round of nodes is terminated.

[0028] In another implementation, it is further determined whether the difference between the next time node and the target arrival point is within a preset threshold. If not, the motion direction corresponding to the next moment of the next moment node is calculated by the opposing resultant potential field, and the pose of the robotic arm platform at the next moment of the next moment node is generated by increasing the motion direction corresponding to the starting node by a preset step size.

[0029] Specifically, the multi-objective optimization of control strategy construction mainly includes the following steps: The system variables are defined as follows: state variables include the robot arm joint angles and angular velocities, and control variables are the joint torques. in, For joint angle vectors, The joint angular velocity vector. This is the joint torque vector.

[0030] The target value function is constructed based on three core indicators: trajectory tracking error, motion smoothness, and system energy consumption. in, , , , which are weighting coefficients used to adjust the relative importance of various performance indicators. , , These represent the objective functions for tracking error, smoothing rate, and energy consumption, respectively.

[0031] Tracking error terms ensures that the robotic arm's end effector moves along a preset cleaning trajectory. The mean square error of the trajectory, with the actual end position as the reference, is expressed as: in, Let i be the position at the end of the i-th time. Let N be the original trajectory point at time i, and N be the number of discrete time steps.

[0032] The smoothness term, which reduces mechanical impact by suppressing abrupt changes in joint acceleration, is expressed as the sum of squares of the second-order differences of joint angles: in, Let be the joint angle at time i. For time step.

[0033] The energy consumption term, represented by the integral of the motor's output function, encourages low-power operation and is expressed as: in, It is the joint torque at the i-th moment under the j-th joint, obtained by the Newton-Euler iteration method. Let be the joint angular velocity at the i-th moment under the j-th joint.

[0034] An initial feasible path is generated by an artificial potential field, and the SQP algorithm is used to iteratively optimize the optimal solution that approaches the target value function, thereby generating the optimal trajectory that satisfies the constraints and makes the target value function converge.

[0035] In step S20, smoothing the initial obstacle avoidance path using a filtering algorithm to obtain a smoothed obstacle avoidance path includes: Set a fitting window, and perform polynomial fitting on each data point in the initial obstacle avoidance path corresponding to the fitting window to obtain a set of polynomials; After analytically solving the set of polynomials, the filter weights corresponding to each data point within the fitting window are calculated to obtain the smoothed trajectory points corresponding to each data point.

[0036] Specifically, a Savitzky-Golay filter is introduced to smooth the path and improve its continuous differentiability; this mainly involves the following steps: Data point polynomial fitting: Within a window of length 2m+1, select data points at each location. The k-th order polynomial fitting is expressed as follows: Among them, the initial trajectory point The fitting coefficients are obtained by fitting using the least squares method. . The filtering process involves analytically solving for the filter weights corresponding to each data point within the window. Thus, the filtering formula is obtained: in, These are the trajectory points obtained after smoothing. The coefficients are obtained from polynomials and applied as fixed weights throughout the trajectory point sequence.

[0037] like Figure 2 As shown, the path smoothing algorithm based on Savitzky-Golay filtering exhibits good adaptability in both circular and rectangular tunnels. After filtering, the motion trajectory of the robotic arm's end effector effectively eliminates the high-frequency jitter and local abrupt changes present in the initial path generated by the artificial potential field method, while maintaining the original path structure. This improvement significantly enhances the stability and control accuracy of trajectory tracking, laying a solid foundation for subsequent control optimization.

[0038] In step 30, setting the dynamic model of the robotic arm platform as a constraint includes: In the dynamic model of the robotic arm platform, the joint torque of the robotic arm in the robotic arm platform is constrained. The joints and angular velocities of the robotic arm in the robotic arm platform do not exceed physical limit constraints; In the robotic arm platform, the distance between the robotic arm and the obstacle is greater than the safe distance threshold constraint.

[0039] Furthermore, the objective value function, which takes tracking error penalty term, smoothness penalty term, and energy consumption penalty term as parameters, includes: The tracking error penalty term is the mean square error between the actual position of the robotic arm end and the reference trajectory when the robotic arm moves along the preset trajectory. The smoothness penalty term is the sum of squares of the second-order differences between the joint angles of the robotic arm; The energy consumption penalty term is characterized by the integral of the dot product output power.

[0040] Specifically, to ensure the safe and stable operation of the robotic arm during coal mine roadway cleaning, and to prevent dangerous operation caused by joint movement, external collisions, or other unavoidable disturbances, multi-dimensional physical constraints need to be applied to its motion control process. Based on the actual needs of the roadway operation scenario and the dynamic characteristics of the robotic arm, this paper designs the following constraints in terms of joint position, joint torque, movement speed, and obstacle avoidance: State constraints: Joint angles and angular velocities must not exceed the physical limits of the robotic arm.

[0041] Control constraints limit joint torque output. ; Obstacle avoidance constraint: the distance between the robotic arm and the obstacle must be greater than a safety threshold. in, The range of joint angular motion and the range of joint torque. , Reachable Joint velocity constraints Reachable , The coordinates of the robotic arm's end effector position. The three-dimensional spatial coordinates of the obstacle The minimum safe distance that the end effector of the robotic arm should maintain from obstacles.

[0042] Furthermore, when constructing the objective value function, for the robotic arm cleaning scenario, a single indicator cannot simultaneously consider accuracy, stability, and economy. This solution constructs a multi-objective value function, based on three core indicators: trajectory tracking error, motion smoothness rate, and system energy consumption. in, , , , which are weighting coefficients used to adjust the relative importance of various performance indicators. , , These represent the objective functions for tracking error, smoothing rate, and energy consumption, respectively.

[0043] Tracking error terms ensures that the robotic arm's end effector moves along a preset cleaning trajectory. The mean square error of the trajectory, with the actual end position as the reference, is expressed as: in, Let i be the position at the end of the i-th time. Let N be the original trajectory point at time i, and N be the number of discrete time steps.

[0044] The smoothness term, which reduces mechanical impact by suppressing abrupt changes in joint acceleration, is expressed as the sum of squares of the second-order differences of joint angles: in, Let be the joint angle at time i. For time step.

[0045] The energy consumption term, represented by the integral of the motor's output function, encourages low-power operation and is expressed as: in, It is the joint torque at the i-th moment under the j-th joint, obtained by the Newton-Euler iteration method. Let be the joint angular velocity at the i-th moment under the j-th joint.

[0046] In step S40, converting the planned path of the robotic arm platform into a continuous motion trajectory, and driving the robotic arm to perform continuous motion by the robotic arm platform control module according to the continuous motion trajectory, includes: The discrete path node sequence in the planned path of the robotic arm platform is converted into a continuous motion trajectory, and corresponding real-time robotic arm joint speed commands and robotic arm acceleration commands are generated. Based on the joint speed command and acceleration command of the robotic arm, the robotic arm platform control module drives the robotic arm to perform continuous movement.

[0047] Specifically, the optimized discrete path node sequence is converted into a continuous motion trajectory, and joint angle, joint velocity, and joint acceleration commands are generated at each moment through time allocation (joint velocity / acceleration constraints). The trajectory commands are then sent to the robotic arm controller to drive the robotic arm to move along the planned path.

[0048] Example 2 This invention also provides a robotic arm platform, such as... Figure 3 As shown, it includes: The initial planning module is used to obtain the current pose of the robotic arm platform, the current environmental information, and the target arrival point. It generates an initial obstacle avoidance path using the gradient descent method based on the adversarial potential field with the current environmental information and the target arrival point as parameters. The path smoothing module is used to smooth the initial obstacle avoidance path using a filtering algorithm to obtain a smoothed obstacle avoidance path. The path optimization module is used to take the dynamic model of the robotic arm platform as a constraint and iterate the objective value function with tracking error penalty, smoothness penalty and energy consumption penalty as parameters to optimize the smoothed obstacle avoidance path, so as to obtain the robotic arm platform planning path corresponding to the convergence of the objective value function. The execution module converts the planned path of the robotic arm platform into a continuous motion trajectory, and the robotic arm platform control module drives the robotic arm to perform continuous motion according to the continuous motion trajectory.

[0049] Real-time obstacle avoidance and path generation are achieved through an artificial potential field, and combined with a multi-objective optimization control algorithm, the relationship between trajectory accuracy, motion stability, and energy efficiency is effectively balanced. This technical solution fully considers the dynamic constraints of the robotic arm and environmental uncertainties, significantly improving the system's operational performance and control quality in confined roadway environments, and providing technical support for achieving efficient and reliable automated cleaning of coal mine roadways.

[0050] Example 3 This invention also provides a computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the robotic arm platform path planning method described in any of the foregoing embodiments.

[0051] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0052] The innovative point that this invention seeks to protect is: (1) A robotic arm trajectory planning method integrating real-time obstacle avoidance using artificial potential field and optimal control using sequential quadratic programming is proposed. The artificial potential field method achieves dynamic obstacle avoidance by constructing a potential field model of the alleyway environment, while the SQP algorithm performs multi-objective optimization of the trajectory under the premise of considering the dynamic constraints of the robotic arm, forming an optimal trajectory that takes into account both obstacle avoidance requirements and motion performance.

[0053] (2) This invention innovatively introduces the Savitzky-Golay filtering algorithm into the path smoothing process, effectively eliminating the high-frequency jitter problem in the path generated by the traditional artificial potential field method. Through local polynomial fitting optimization, the smoothness and trackability of the trajectory are significantly improved while maintaining the path topology, laying the foundation for high-precision trajectory control.

[0054] (3) A multi-objective value function containing tracking error penalty, smoothness penalty, and energy consumption penalty was constructed and solved using the SQP algorithm in the fmincon optimization toolbox. The value function balances the performance indicators through weighted fusion and verifies its comprehensive advantages in trajectory accuracy, motion stability, and energy efficiency in a real coal mine roadway experimental environment.

[0055] (4) The method was systematically verified based on the actual coal mine roadway scenario of a large domestic energy group. The results show that the method has excellent engineering applicability in complex underground environments. Compared with traditional planning methods, it shows significant improvements in key indicators such as trajectory tracking error, motion smoothness and system energy consumption, providing a reliable technical solution for the automation upgrade of coal mine roadway cleaning operations.

[0056] Experimental results This invention focuses on trajectory optimization, constructing a nonlinear programming (NLP) problem by combining constraints and a robotic arm dynamics model to achieve path optimization. For solving the NLP problem, the `fmincon` function in MATLAB, based on the SQP algorithm, is selected. This function gradually approaches the optimal solution of the objective function, ultimately generating the optimal trajectory that meets the task requirements.

[0057] To further demonstrate the practical application effect of the cleaning system in complex downhole environments, the weight coefficients of each sub-objective in the multi-objective function were adjusted in a roadway environment. Meanwhile, to ensure the stable convergence and computational efficiency of the SQP algorithm, key optimization parameters such as the maximum number of iterations and the convergence threshold for step size variation were set. The specific settings for this simulation are detailed in Table 1.

[0058] Table 1 Weighting coefficients and optimization parameter settings Based on the above parameter configuration, the comparison of joint angle change curves obtained under the fmincon solver iteration is shown in the figure below. Figure 4 The figure shows a comparison of the joint angle curves before and after optimization, obtained using the fmincon solver and SQP algorithm in a rectangular tunnel environment. The blue dashed line represents the initial joint angle change curve, while the red solid line represents the optimized joint angle change curve. As can be seen from the figure, the optimized joint angle curve exhibits higher smoothness, eliminating some high-frequency oscillations and unnecessary violent fluctuations, effectively improving the continuity and controllability of the trajectory, and contributing to the improvement of system execution stability and the service life of the mechanical structure.

[0059] To evaluate the optimization effect of trajectory optimization on dynamic characteristics in a rectangular tunnel environment, a comparison graph of joint angular velocity changes before and after optimization was plotted, as shown below. Figure 5 As shown, the optimization effect on angular acceleration before and after trajectory optimization is analyzed. A comparison of the joint angular velocity change curves before and after trajectory optimization is presented in a rectangular tunnel environment. It can be observed that the angular velocity curve before optimization fluctuates significantly, with frequent and drastic velocity changes, which not only increases the burden on the control system but may also cause actuator jitter or control instability. In contrast, the optimized joint angular velocity curve exhibits higher smoothness, significantly reduces the fluctuation amplitude of the curve, and the change curve is more continuous, demonstrating superior dynamic response characteristics. Effective optimization of dynamic performance is achieved while maintaining trajectory tracking accuracy.

[0060] To further evaluate the optimization effect of trajectory optimization on dynamic characteristics, a comparison chart of joint angular acceleration changes before and after optimization in a rectangular tunnel environment was plotted, as shown below. Figure 6 The image shows a comparison of the angular acceleration of each joint of the robotic arm before and after optimization in a rectangular tunnel environment. Before optimization, the angular acceleration curve fluctuated violently, frequently switching between positive and negative values, and was accompanied by multiple spike pulse oscillations, resulting in poor continuity. After optimization, the curve became significantly smoother and remained basically stable throughout the entire process. The optimized control strategy significantly improved the abrupt changes in the original trajectory, exhibiting a smoother dynamic response. Simultaneously, the optimized strategy reduced instantaneous impact forces on the joints, decreased wear on the mechanism, and helped extend the lifespan of the mechanical structure. The optimized results effectively improved the dynamic performance of the robotic arm, laying a solid algorithmic theoretical foundation for the subsequent deployment of the control system.

[0061] This invention performs trajectory tracking simulation analysis on a six-degree-of-freedom robotic arm under no-load conditions. By performing dynamic solutions on the optimized joint angle sequence, the curves of the required driving torque changes for each joint during the execution of the robotic arm are plotted, such as... Figure 7 As shown in the figure, this graph visually reflects the magnitude and trend of the torque required by each joint to complete the desired trajectory over different time periods.

[0062] like Figure 7 As shown, the optimized trajectory exhibits relatively stable torque output at each joint during execution, without significant abrupt changes or drastic fluctuations, demonstrating high stability. During critical phases of the task, such as large rotations or near path turning points, each joint maintains minimal torque fluctuations, showcasing good mechanical coordination and control robustness. Compared to the maximum allowable joint torque of the actuator, its value is within a reasonable range, effectively avoiding the risk of actuator overload while meeting the dynamic requirements of trajectory tracking.

[0063] Verification and analysis of on-site operation test results: This study selected a coal mine roadway actually operated by a large domestic energy group as the verification scenario, and conducted a comprehensive test of the cleaning robotic arm system for more than 20 hours under real operating conditions. This experimental scenario fully reflects the characteristics of complex underground working conditions, including irregular roadway structures, high-intensity coal dust environments, and typical obstacle distributions. This makes the experimental results representative of the industry, technologically advanced, and valuable for engineering application, providing a reliable basis for the application of automated cleaning technology in similar mines.

[0064] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A path planning method for a robotic arm platform, characterized in that, include: The current pose, current environment information, and target arrival point of the robotic arm platform are obtained. The initial obstacle avoidance path is generated by using the gradient descent method with the current environment information and the target arrival point as parameters to generate the counteracting resultant potential field. The initial obstacle avoidance path is smoothed using a filtering algorithm to obtain a smoothed obstacle avoidance path. The dynamic model of the robotic arm platform is used as a constraint, and the objective value function with tracking error penalty, smoothness penalty and energy consumption penalty as parameters is iterated to optimize the smoothed obstacle avoidance path, so as to obtain the robotic arm platform planning path corresponding to the convergence of the objective value function. The planned path of the robotic arm platform is converted into a continuous motion trajectory, and the robotic arm is driven to perform continuous motion by the robotic arm platform control module according to the continuous motion trajectory.

2. The path planning method for a robotic arm platform according to claim 1, characterized in that, The counteracting resultant potential field, which uses the current environmental information and the target arrival point as parameters, includes: Using the target arrival point as a reference, the attraction potential energy is used to solve the quadratic function composed of the starting point of the robotic arm and the target arrival point of the robotic arm, so as to obtain the attraction potential field of the target arrival point of the robotic arm. The repulsive potential energy of the obstacle information on the robot arm in the current environment is taken as a reference. The inverse proportional function established with the shortest distance from the robot arm to the obstacle and the range of the repulsive force of the obstacle on the robot arm is solved to obtain the repulsive potential field of the obstacle. The attractive potential field at the target arrival point of the robotic arm and the repulsive potential field of the obstacle are combined to obtain the counteracting resultant potential field.

3. The path planning method for a robotic arm platform according to claim 1, characterized in that, The step of generating an initial obstacle avoidance path using the gradient descent method based on the adversarial potential field with the current environmental information and the target arrival point as parameters includes: The current pose of the robotic arm platform is used as the starting node, and the motion direction corresponding to the starting node is calculated through the opposing resultant potential field. The next pose of the robotic arm platform is generated by moving along the direction corresponding to the starting node and increasing by a preset step size. Determine whether the next pose of the robotic arm platform will collide with an obstacle; If there is no collision, the next pose of the robotic arm platform is added to the path node to obtain the next moment node, and a termination judgment is made.

4. The path planning method for a robotic arm platform according to claim 1, characterized in that, The step of generating an initial obstacle avoidance path using the gradient descent method with the current environmental information and the target arrival point as parameters, against the resultant potential field, includes: The current pose of the robotic arm platform is used as the starting node, and the motion direction corresponding to the starting node is calculated through the opposing resultant potential field. The next pose of the robotic arm platform is generated by moving along the direction corresponding to the starting node and increasing by a preset step size. Determine whether the next pose of the robotic arm platform will collide with an obstacle; If a collision occurs, the preset step size is reduced, the next pose of the robotic arm platform is regenerated, and it is re-determined whether the next pose of the robotic arm platform will collide with an obstacle.

5. The path planning method for a robotic arm platform according to claim 1, characterized in that, The constraint condition for using the dynamic model of the robotic arm platform includes: In the dynamic model of the robotic arm platform, the joint torque of the robotic arm in the robotic arm platform is constrained. The joints and angular velocities of the robotic arm in the robotic arm platform do not exceed physical limit constraints; In the robotic arm platform, the distance between the robotic arm and the obstacle is greater than the safe distance threshold constraint.

6. The path planning method for a robotic arm platform according to claim 1, characterized in that, The step of converting the planned path of the robotic arm platform into a continuous motion trajectory, and driving the robotic arm to perform continuous motion by the robotic arm platform control module according to the continuous motion trajectory, includes: The discrete path node sequence in the planned path of the robotic arm platform is converted into a continuous motion trajectory, and corresponding real-time robotic arm joint speed commands and robotic arm acceleration commands are generated. Based on the joint speed command and acceleration command of the robotic arm, the robotic arm platform control module drives the robotic arm to perform continuous movement.

7. A robotic arm platform, characterized in that, include: The initial planning module is used to obtain the current pose of the robotic arm platform, the current environmental information, and the target arrival point. It generates an initial obstacle avoidance path using the gradient descent method based on the adversarial potential field with the current environmental information and the target arrival point as parameters. The path smoothing module is used to smooth the initial obstacle avoidance path using a filtering algorithm to obtain a smoothed obstacle avoidance path. The path optimization module is used to take the dynamic model of the robotic arm platform as a constraint and iterate the objective value function with tracking error penalty, smoothness penalty and energy consumption penalty as parameters to optimize the smoothed obstacle avoidance path, so as to obtain the robotic arm platform planning path corresponding to the convergence of the objective value function. The execution module converts the planned path of the robotic arm platform into a continuous motion trajectory, and the robotic arm platform control module drives the robotic arm to perform continuous motion according to the continuous motion trajectory.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs that can be executed by one or more processors to implement the robotic arm platform path planning method according to any one of claims 1 to 6.

Images