Trajectory planning method for redundant robot arm

Abstract

This invention relates to a trajectory planning method for a redundant robotic arm. The method divides the redundant robotic arm into two groups of two, with the group containing the end effector being the first group. All other groups mimic the motion of the first group. By planning the trajectory of the first group to avoid obstacles, the motion trajectories of the remaining groups are determined. This invention not only reduces the computation of trajectory control schemes for the remaining parts of the redundant robotic arm but also ensures that each component of the redundant robotic arm avoids obstacles. Furthermore, the process of the first group sequentially reaching each segment is treated as an independent process; therefore, this invention can independently and simultaneously calculate the obstacle-avoidance trajectory for each part of the configuration.

Description

Technical Field

[0001] This invention belongs to the field of robot trajectory design and relates to a trajectory planning method for a redundant robotic arm. Background Technology

[0002] Redundant robotic arms, with their greater degrees of freedom compared to traditional robotic arms, have become a research hotspot in recent years due to their high flexibility, adaptability, obstacle avoidance, singularity avoidance, and joint constraint avoidance capabilities. They are currently widely used in industry, medical services, and aerospace. However, because redundant robotic arms have more degrees of freedom, there are countless solutions for their end effector to reach a designated target position from its initial position. Furthermore, the presence of obstacles in space means that many of these solutions are invalid, leading to collisions between the robotic arm and obstacles. Therefore, trajectory planning for redundant robotic arms becomes extremely difficult.

[0003] Trajectory planning is a fundamental problem in the motion control of robotic arms, and the result of trajectory planning can measure whether the robot has performed the task well. Currently, most trajectory planning methods for robotic arms calculate the trajectory using polynomial interpolation based on specific constraints such as velocity and acceleration in joints or Cartesian space. The trajectory planning problem in Cartesian space is a key issue for redundant robotic arms performing complex tasks. Cartesian space trajectory planning refers to inversely solving the trajectory planned by the robotic arm's end effector from the starting point to the ending point into joint space for each point.

[0004] Cartesian space trajectory planning is one of the main methods in common robotic arm trajectory planning schemes. Cartesian space trajectory planning refers to planning the motion trajectory of the robotic arm's end effector in a Cartesian coordinate system. This typically involves determining the position and orientation of the end effector in three-dimensional space, and then implementing the robotic arm's motion through a series of path planning algorithms. Redundant robotic arms often have multiple joint angle solutions when reaching the same target position and orientation, meaning multiple solutions exist. When the end effector of a redundant robotic arm avoids obstacles and reaches the designated orientation, the intermediate components of the robotic arm become uncontrolled and prone to collisions with obstacles.

[0005] The collision-free fast path planning algorithm for hyperredundant robotic arms based on reinforcement learning is another commonly used robotic arm planning method. This method utilizes the obstacle-avoiding movement pattern of a snake crawling, meaning that only the movement path of the end effector needs to be planned, and the rest of the body follows the end effector. For example... Figure 1 As shown, it uses attitude interpolation to adjust the first arm, which has not yet entered the configuration, to be collinear with the attitude indicated by the last line segment of the configuration diagram. Simultaneously, as the arm enters, the attitude of the arm already conforming to the configuration is adjusted to conform to the next attitude. For example... Figure 2As shown, because the links of the hyper-redundant robotic arm are made of rigid materials, it cannot achieve a snake-like level of flexibility during movement; that is, the arm's body cannot always align with the path of the end effector. Furthermore, since the arm, which has already partially conformed to the configuration diagram, is randomly adjusted to fit the next posture, the offset distance accumulates as the robotic arm's degrees of freedom increase. Simultaneously, as the degrees of freedom of the redundant robotic arm increase, the space required for the end effector to move to a specified posture also increases. The previous and subsequent motion trajectories of the redundant robotic arm are strongly coupled; the trajectory of the previous posture can only be determined before finding the trajectory scheme for the next posture. This also limits the application of collision-free fast path planning algorithms for hyper-redundant robotic arms based on reinforcement learning. Summary of the Invention

[0006] The technical problem solved by this invention is to overcome the shortcomings of the prior art and propose a trajectory planning method for redundant robotic arms. This method can effectively reduce the number of low-quality solutions when the redundant robotic arm reaches a specified posture, help all components of the redundant robotic arm avoid obstacles, and independently and simultaneously calculate the obstacle avoidance trajectory of each part of the configuration.

[0007] The solution of the present invention is:

[0008] A trajectory planning method for a redundant robotic arm includes:

[0009] Step 1: Divide the redundant robotic arm into motion groups. Starting from the end of the robotic arm, each pair of arms and the three joints connected to those two arms form a group. Starting from the end of the robotic arm and moving forward sequentially, name the group containing the end effector as the first group, then the second group, the third group, and so on, up to the Nth group.

[0010] Step 2: Determine the configuration of the robotic arm end effector when it reaches the designated target position. The configuration is the posture of the redundant robotic arm. When the redundant robotic arm reaches this posture, none of its components collide with obstacles.

[0011] Step 3: Add two horizontal virtual attitude segments at the very end of the configuration. Starting from the last virtual attitude segment, extract several segments by taking every three consecutive line segments as a fragment.

[0012] Step 4: Name the first and last joints of the first group as joint I and joint II respectively. During the movement of the first group, restrict the trajectories of joint I and joint II to be on the line segments of the configuration. The trajectory of joint II of the first group is consistent with the trajectory of joint I when the first group reaches the previous posture.

[0013] Step 5: Select the trajectory movement plan of the first group on each segment, and calculate the deviation distance between the first group and each segment. Based on the deviation distance, determine whether the first group collides with the obstacle at some segments. If it does, proceed to Step 6; if no collision occurs in any segment, find the full trajectory movement plan of the first group and proceed to Step 7.

[0014] Step Six: Change the trajectory movement scheme of the first group on the collision segment, and determine whether the first group collides with the obstacle at the corresponding segment. If no collision occurs, find the full trajectory movement scheme of the first group and proceed to Step Seven; otherwise, continue to change the trajectory movement scheme of the first group on the collision segment. If there is still a collision after 20 times, return to Step Two and change the configuration.

[0015] Step 7: The trajectory of group n is consistent with the trajectory of group n-1 in the previous pose, n = 2, 3, ... N, thus completing the trajectory planning.

[0016] Preferably, the deviation distance refers to the vertical distance from the geometric center of the intermediate joint of the first group to the current segment.

[0017] Preferably, the end effector is regarded as a joint.

[0018] Preferably, two adjacent groups share a joint.

[0019] A trajectory planning system for a redundant robotic arm includes a robotic arm partitioning module, a configuration determination module, a configuration improvement module, a trajectory constraint module, a first group trajectory determination module, and a robotic arm trajectory planning module.

[0020] Robotic arm partitioning module: Divide redundant robotic arms into motion groups, starting from the end of the robotic arm, each group consists of two arms and three joints connected to those two arms; moving forward sequentially from the end of the robotic arm, the group containing the end effector is named the first group, followed by the second group, the third group, ... the Nth group;

[0021] Configuration determination module: Determines the configuration of the robotic arm end effector when it reaches the specified target position. The configuration is the posture of the redundant robotic arm. When the redundant robotic arm reaches this posture, none of its components collide with obstacles.

[0022] Configuration improvement module: Add two horizontal virtual attitude segments at the very end of the configuration. Starting from the last virtual attitude segment, extract several segments by taking every three consecutive line segments as a fragment.

[0023] Trajectory restriction module: The first and last joints of the first group are named joint I and joint II respectively. During the movement of the first group, the trajectories of joint I and joint II are restricted to be on the line segments of the configuration. The trajectory of joint II of the first group is consistent with the trajectory of joint I when the first group reaches the previous posture.

[0024] The first group trajectory determination module selects the trajectory movement scheme of the first group on each segment and calculates the deviation distance between the first group and each segment. Based on the deviation distance, it determines whether the first group collides with the obstacle at some segments. If no collision occurs in any segment, the entire trajectory movement scheme of the first group is found. If a collision occurs, the trajectory movement scheme of the first group on the segment where the collision occurred is changed, and it is determined whether the first group collides with the obstacle at the corresponding segment. If no collision occurs, the entire trajectory movement scheme of the first group is found. Otherwise, the trajectory movement scheme of the first group on the segment where the collision occurred is changed again. If a collision still occurs after 20 iterations, the module returns to the configuration improvement module to change the configuration.

[0025] Robotic arm trajectory planning module: controls the trajectory of the nth group to be consistent with the trajectory of the (n-1)th group in the previous posture, n = 2, 3, ... N.

[0026] Preferably, in the first group trajectory determination module, the deviation distance refers to the vertical distance from the geometric center of the intermediate joint of the first group to the current segment.

[0027] Preferably, in the robotic arm segmentation module, the end effector is regarded as a joint.

[0028] Preferably, in the robotic arm division module, two adjacent groups share a joint.

[0029] The advantages of this invention compared to the prior art are:

[0030] In this invention, after the trajectory planning of the first group is completed, the remaining groups of the redundant robotic arm completely mimic the trajectory movement of the first group during its posture changes. The deviation distance of the remaining groups during each posture change process is exactly the same as that of the first group. This not only reduces the calculation of trajectory control schemes for the remaining parts of the redundant robotic arm, but also ensures that each component of the redundant robotic arm avoids obstacles. Furthermore, the process of the first group sequentially reaching each segment is considered an independent process; therefore, this invention can independently and simultaneously calculate the obstacle-avoiding trajectory for each part of the configuration's posture. Attached Figure Description

[0031] Figure 1 The motion control graph of the robotic arm using a reinforcement learning-based hyper-redundant robotic arm collision-free fast path planning algorithm;

[0032] Figure 2 A schematic diagram of the offset distance using a collision-free fast path planning algorithm for an ultra-redundant robotic arm based on reinforcement learning;

[0033] Figure 3 This is a schematic diagram of the robotic arm being divided into groups according to the present invention;

[0034] Figure 4 This is a configuration diagram of the present invention;

[0035] Figure 5 The first group's motion diagram;

[0036] Figure 6 This is a deviation distance diagram of the present invention;

[0037] Figure 7 The trajectory diagram is for a collision-free fast path planning algorithm for a super-redundant robotic arm based on reinforcement learning.

[0038] Figure 8 A trajectory diagram using the method of the present invention;

[0039] Figure 9 Construction graph for adding virtual poses;

[0040] Figure 10 This is a fragment of the constructed graph. Detailed Implementation

[0041] The invention will now be further described with reference to the accompanying drawings.

[0042] This invention proposes a novel trajectory planning method for redundant robotic arms. By dividing the redundant robotic arms into two groups, the first group includes the end effector, and all other groups mimic the movement of the first group. The method seeks to find the trajectory planning of all parts of the movement group that avoid obstacles, and specifies the movement trajectory of the remaining movement groups.

[0043] Specifically, the specific solution of the present invention is as follows:

[0044] 1. Classification of redundant robotic arms

[0045] Figure 3 This is a schematic diagram illustrating the grouping of a 6-DOF robotic arm. First, the redundant robotic arm is divided into groups, starting from the end effector. Each group consists of two arms and three joints, with the end effector also considered a special joint. Adjacent groups share a joint. The group containing the end effector is named Group 1, followed by Group 2, Group 3, and so on.

[0046] 2. Trajectory Constraints

[0047] Taking a redundant robotic arm with 10 degrees of freedom as an example, Figure 4For a configuration diagram of a 10-DOF robotic arm found through searching, the configuration diagram is considered to be composed of ten line segments a, b, c, d, e, f, g, h, i, and j. The configuration diagram determines the posture of a redundant robotic arm. When the redundant robotic arm reaches this posture, the end effector of the robotic arm reaches the specified target position, and at this time, none of the components of the redundant robotic arm collide with obstacles.

[0048] First, focus only on the trajectory of the first group. For example... Figure 5 As shown, assume the first group has already moved to Figure 4 The posture shown by line segments a and b in the configuration diagram will next move to the posture shown by line segments b and c. The head and tail joints of the first group are named Joint I and Joint II, respectively. During the above posture changes of the first group, Joint I is restricted to move along line segment c, and Joint II is restricted to move along line segment a. This completes the trajectory constraint conditions.

[0049] 3. Calculate the deviation distance

[0050] Although the end effector reaches the target position when the redundant robotic arm reaches the posture specified in the configuration diagram, and none of the components of the redundant robotic arm collide with obstacles at this point, collisions with obstacles are still possible during the process of the redundant robotic arm moving from the initial posture to the final specified posture. For example... Figure 6 As shown, during the movement, the vertical distance from the geometric center of the intermediate joint of the robotic arm group to the configuration is called the deviation distance.

[0051] like Figure 7 As shown, when using the reinforcement learning-based hyperredundant robotic arm collision-free fast path planning algorithm mentioned in the background section, the control of the robotic arm from... Figure 4 During the movement from the postures specified by line segments a, b, c, d, e, f in the configuration diagram to the postures specified by line segments b, c, d, e, f, g, the deviation distance increases sharply as the degrees of freedom of the robotic arm increase. This is because all arms of the redundant robotic arm mimic the trajectory of the foremost arm. However, the foremost arm needs to sequentially reach the postures indicated by line segments a, b, c, d, e, f, g, h, i, j from its initial position. During the movement to a certain posture, the trajectory of the arm's rear end is consistent with the trajectory of its front end when it reached the previous posture. Therefore, only by determining the trajectory to reach the previous posture can the trajectory to reach the next posture be found.

[0052] like Figure 8 As shown, using the method of the present invention, the robotic arm is controlled to move from... Figure 4During the movement from the postures specified by line segments a, b, c, d, e, f in the configuration diagram to the postures specified by line segments b, c, d, e, f, g, the deviation distance of the robotic arm is significantly smaller than that achieved using a reinforcement learning-based hyper-redundant robotic arm collision-free fast path planning algorithm. The first group needs to sequentially reach the postures indicated by line segments a, b, c, d, e, f, g, h, i, j from its initial position. During the movement to a certain posture, the trajectory of joint II in the first group is consistent with the trajectory of joint I when the first group reaches the previous posture. The method of this invention restricts the trajectories of joint I and joint II to be on the line segments of the configuration. This not only reduces the number of solutions where the redundant robotic arm collides with obstacles but also allows for separate searching of the trajectory to reach each posture.

[0053] like Figure 9 As shown, for a redundant robotic arm with 10 degrees of freedom, a condition is determined as follows: Figure 4 After the configuration is completed, at the position of line segment 'a' at the very end of the configuration, two horizontal virtual poses are added, corresponding to line segment virtual 1 and virtual 2 respectively. For example... Figure 10 As shown, Figure 9 The configuration in the image was extracted according to the poses corresponding to three consecutive line segments, and 10 segments were extracted from each segment.

[0054] Since the trajectories of joints I and II in the first group are restricted to moving along the configuration, the process of the first group arriving at each segment sequentially can be considered as an independent process. Assume the first group has already arrived... Figure 10 The pose indicated by the last two line segments out of the 10 segments is the pose that the movement will move to, for example, the pose specified by line segments virtual 1 and virtual 2 in the first segment, and the pose specified by line segments virtual 2 and 'a'. Simultaneously calculate... Figure 10 The minimum deviation distance of 10 segments. If the deviation distance of each segment ensures that the first group does not collide with obstacles, then a full trajectory control scheme for the first group is found. Figure 10 If the first group of 10 segments collides with an obstacle due to a deviation distance control in one segment, the trajectory movement scheme of the first group on the segment where the collision occurred is changed. It is determined whether the first group and the obstacle collided at the corresponding segment. If no collision occurred, the full trajectory movement scheme of the first group is found; otherwise, the trajectory movement scheme of the first group on the segment where the collision occurred is changed again. If a segment still collides after 20 times, the configuration is changed.

[0055] 3. Trajectory Control Method

[0056] Once the trajectory control of the first group has been established, the remaining groups of the redundant robotic arm completely mimic the trajectory movement of the first group during its attitude changes. The deviation distance of the remaining groups during each attitude change is exactly the same as that of the first group. This not only reduces the calculation of trajectory control schemes for the remaining parts of the redundant robotic arm, but also ensures that all components of the redundant robotic arm avoid obstacles.

[0057] The present invention also provides a trajectory planning system for a redundant robotic arm, including a robotic arm partitioning module, a configuration determination module, a configuration improvement module, a trajectory restriction module, a first group trajectory determination module, and a robotic arm trajectory planning module.

[0058] Robotic arm partitioning module: Redundant robotic arms are divided into motion groups. Starting from the end effector of the robotic arm, every two arms and the three joints connected to those two arms form a group; the end effector is regarded as a joint, and adjacent groups share a joint. Starting from the end effector of the robotic arm and moving forward sequentially, the group containing the end effector is named the first group, followed by the second group, the third group, ... the Nth group.

[0059] Configuration determination module: Determines the configuration of the robotic arm end effector when it reaches the specified target position. The configuration is the posture of the redundant robotic arm. When the redundant robotic arm reaches this posture, none of its components collide with obstacles.

[0060] Configuration improvement module: Add two horizontal virtual attitude segments at the very end of the configuration. Starting from the last virtual attitude segment, extract several segments by taking every three consecutive line segments as a fragment.

[0061] Trajectory restriction module: The first and last joints of the first group are named joint I and joint II respectively. During the movement of the first group, the trajectories of joint I and joint II are restricted to be on the line segments of the configuration. The trajectory of joint II of the first group is consistent with the trajectory of joint I when the first group reaches the previous posture.

[0062] The first group trajectory determination module selects the trajectory motion scheme of the first group on each segment and calculates the deviation distance between the first group and each segment. Based on the deviation distance, it determines whether the first group collides with obstacles at some segments. If no collision occurs in any segments, the full trajectory motion scheme of the first group is found. If a collision occurs, the trajectory motion scheme of the first group on the segment where the collision occurred is changed, and it is determined whether the first group collides with obstacles at the corresponding segment. If no collision occurs, the full trajectory motion scheme of the first group is found; otherwise, the trajectory motion scheme of the first group on the segment where the collision occurred continues to be changed. If collisions still occur after 20 iterations, the module returns to the configuration improvement module to change the configuration. The deviation distance refers to the vertical distance from the geometric center of the first group's intermediate joint to the current segment.

[0063] Robotic arm trajectory planning module: controls the trajectory of the nth group to be consistent with the trajectory of the (n-1)th group in the previous posture, n = 2, 3, ... N.

[0064] This invention effectively reduces the number of low-quality solutions when a redundant robotic arm reaches a specified posture, helps all components of the redundant robotic arm avoid obstacles, and can independently calculate whether there is a trajectory solution that can avoid obstacles for each posture of the configuration.

[0065] The contents not described in detail in this specification are common knowledge to those skilled in the art.

Claims

1. A trajectory planning method for a redundant robotic arm, characterized in that, Includes the following steps: Step 1: Divide the redundant robotic arm into motion groups. Starting from the end of the robotic arm, each pair of arms and the three joints connected to those two arms form a group. Starting from the end of the robotic arm and moving forward sequentially, name the group containing the end effector as the first group, then the second group, the third group, and so on, up to the Nth group. Step 2: Determine the configuration of the robotic arm end effector when it reaches the designated target position. The configuration is the posture of the redundant robotic arm. When the redundant robotic arm reaches this posture, none of its components collide with obstacles. Step 3: Add two horizontal virtual attitude segments at the very end of the configuration. Starting from the last virtual attitude segment, extract several segments by taking every three consecutive line segments as a fragment. Step 4: Name the first and last joints of the first group as joint I and joint II respectively. During the movement of the first group, restrict the trajectories of joint I and joint II to be on the line segments of the configuration. The trajectory of joint II of the first group is consistent with the trajectory of joint I when the first group reaches the previous posture. Step 5: Select the trajectory movement plan of the first group on each segment, and calculate the deviation distance between the first group and each segment. Based on the deviation distance, determine whether the first group collides with the obstacle at some segments. If it does, proceed to Step 6; if no collision occurs in any segment, find the full trajectory movement plan of the first group and proceed to Step 7. Step 6: Change the trajectory of the first group on the collision segment and determine whether the first group collides with the obstacle at the corresponding segment. If no collision occurs, find the full trajectory of the first group and proceed to Step 7; otherwise, continue to change the trajectory of the first group on the collision segment. If collision still occurs after 20 attempts, return to Step 2 and change the configuration. Step 7: The trajectory of group n is consistent with the trajectory of group n-1 in the previous pose, n = 2, 3, ... N, thus completing the trajectory planning.

2. The trajectory planning method for a redundant robotic arm according to claim 1, characterized in that, Deviation distance refers to the vertical distance from the geometric center of the intermediate joint of the first group to the current segment.

3. The trajectory planning method for a redundant robotic arm according to claim 1, characterized in that, The end effector is considered as a joint.

4. The trajectory planning method for a redundant robotic arm according to claim 1, characterized in that, Two adjacent groups share a joint.

5. A trajectory planning system for a redundant robotic arm, characterized in that, It includes a robotic arm segmentation module, a configuration determination module, a configuration improvement module, a trajectory constraint module, a first group trajectory determination module, and a robotic arm trajectory planning module; Robotic arm partitioning module: Divide redundant robotic arms into motion groups, starting from the end of the robotic arm, each group consists of two arms and three joints connected to those two arms; moving forward sequentially from the end of the robotic arm, the group containing the end effector is named the first group, followed by the second group, the third group, ... the Nth group; Configuration determination module: Determines the configuration of the robotic arm end effector when it reaches the specified target position. The configuration is the posture of the redundant robotic arm. When the redundant robotic arm reaches this posture, none of its components collide with obstacles. Configuration improvement module: Add two horizontal virtual attitude segments at the very end of the configuration. Starting from the last virtual attitude segment, extract several segments by taking every three consecutive line segments as a fragment. Trajectory restriction module: The first and last joints of the first group are named joint I and joint II respectively. During the movement of the first group, the trajectories of joint I and joint II are restricted to be on the line segments of the configuration. The trajectory of joint II of the first group is consistent with the trajectory of joint I when the first group reaches the previous posture. The first group trajectory determination module selects the trajectory movement scheme of the first group on each segment and calculates the deviation distance between the first group and each segment. Based on the deviation distance, it determines whether the first group collides with the obstacle at some segments. If no collision occurs in any segment, the entire trajectory movement scheme of the first group is found. If a collision occurs, the trajectory movement scheme of the first group on the segment where the collision occurred is changed, and it is determined whether the first group collides with the obstacle at the corresponding segment. If no collision occurs, the entire trajectory movement scheme of the first group is found. Otherwise, the trajectory movement scheme of the first group on the segment where the collision occurred is changed again. If a collision still occurs after 20 iterations, the module returns to the configuration improvement module to change the configuration. Robotic arm trajectory planning module: controls the trajectory of the nth group to be consistent with the trajectory of the (n-1)th group in the previous posture, n = 2, 3, ... N.

6. The trajectory planning system for a redundant robotic arm according to claim 5, characterized in that, In the first group trajectory determination module, the deviation distance refers to the vertical distance from the geometric center of the middle joint of the first group to the current segment.

7. The trajectory planning system for a redundant robotic arm according to claim 5, characterized in that, In the robotic arm segmentation module, the end effector is regarded as a joint.

8. The trajectory planning system for a redundant robotic arm according to claim 5, characterized in that, In the robotic arm segmentation module, adjacent groups share a joint.

Images