Automatic obstacle avoidance method for mobile manipulator based on null-space control and intention guidance

Abstract

The invention discloses a null space control and intention guidance-based mobile mechanical arm autonomous obstacle avoidance method, which comprises the following steps of: constructing a multi-layer local environment sensing point cloud map through point cloud data, generating a multi-layer sensing range from inside to outside, and reducing the sensitivity of the outer-layer sensing range; the lateral obstacle avoidance linear speed of the mobile chassis is generated by determining the obstacle avoidance emergency degree, the multi-level local environment perception and the point cloud map of the desensitization area, and the final lateral obstacle avoidance speed accepted by the mobile chassis is obtained by combining the maximum lateral movement speed allowed by the mobile chassis; the speed direction of projection of an end effector on the ground is defined by moving the end effector of the mechanical arm, the compensation gain coefficient and the chassis deflection angle are combined to generate the movement intention following compensation angular speed of the collaborator, and the movement intention following speed of the collaborator is obtained through the compensation angular speed and the lateral obstacle avoidance speed; and a kinematic model of the mobile mechanical arm is constructed, and autonomous obstacle avoidance of the mobile chassis with man-machine cooperation is achieved by combining the movement intention of the collaborator with the speed.

Description

Technical Field

[0001] The present invention belongs to the technical field of robot motion control, and particularly relates to a method for autonomous obstacle avoidance of a mobile manipulator based on null space control and intention guidance. Background Art

[0002] Integrating capabilities such as perception, operation, and movement has become a research hotspot and development direction in robotics technology in recent years. A mobile manipulator composed of a manipulator and a mobile platform has both flexible movement and dexterous operation capabilities, and it is gradually applied in industrial, agricultural, and service fields, etc., with broad application prospects. However, the strong non-linear and complex dynamic coupling factors between the mobile platform and the manipulator directly affect the motion control performance of the mobile manipulator. How to develop an efficient and advanced autonomous motion controller is still the main challenge to demonstrate its application potential. In the scenario of human-robot collaboration, how to consider the differences in the motion characteristics of the mobile platform and the manipulator, and coordinate redundant degrees of freedom to achieve autonomous obstacle avoidance and real-time motion control in an unstructured environment is the key challenge faced in its human-robot collaborative operation.

[0003] Traditional coordinated motion control methods for mobile manipulators mostly focus on decoupling control strategies, that is, using an upper-level planner to directly allocate task objectives for the mobile platform and the manipulator, and then designing independent controllers for the two respectively to complete job control. It can be seen that this upper-level planner cannot achieve the coordinated allocation of the full-body redundant degrees of freedom for multi-task synchronization of the mobile manipulator, and can only be used for discrete sequential job control of the mobile manipulator. Another research direction for coordinated motion control of mobile manipulators focuses on joint control strategies, that is, by solving the full-body redundant kinematics of the mobile manipulator to complete the primary task and the secondary task simultaneously, which can meet the requirements of vehicle-arm coordinated motion control for continuous operation of the mobile manipulator. Currently, methods such as multi-task hierarchical control and weighted least squares method have been applied to the redundant kinematic coordinated control of mobile manipulators. These methods optimize specific indicators in a gradient descent manner, but the gradient optimization method is difficult to control the optimization degree and quantify the optimization indicators, and does not consider the autonomous obstacle avoidance problem of the mobile chassis. In terms of obstacle avoidance of the mobile chassis, existing research is limited to the global obstacle avoidance and local obstacle avoidance algorithms of the mobile chassis itself, lacking the utilization of the overall motion information of the mobile manipulator, resulting in a large computational cost for its obstacle avoidance algorithm, poor obstacle avoidance effect, and lack of consideration for the interaction experience of collaborators. Summary of the Invention

[0004] To solve the deficiencies of the prior art and achieve the purpose of autonomous obstacle avoidance of a mobile manipulator with null space control and human intention guidance, the present invention adopts the following technical solutions:

[0005] A method for autonomous obstacle avoidance of a mobile manipulator based on null space control and intention guidance, comprising the following steps:

[0006] Step 1: Obtain point cloud data and construct a multi-level local environment perception point cloud map. Based on the multi-level perception range from the inside out of the point cloud map, reduce the sensitivity of the outer perception range to meet the actual obstacle avoidance requirements in the human-robot collaboration scenario;

[0007] Step 2: Generate a lateral obstacle avoidance speed command and a collaborator motion intention following speed command to meet the obstacle avoidance requirements of the mobile chassis in an unstructured scenario; The generation of the lateral obstacle avoidance speed command is achieved by defining the obstacle avoidance weights of each layer of the perception area to determine the urgency of obstacle avoidance from the inside out. According to the multi-level local environment perception and the desensitized area point cloud map, generate the lateral obstacle avoidance linear speed of the mobile chassis. Combining with the maximum lateral movement speed allowed by the mobile chassis, obtain the final lateral obstacle avoidance speed accepted by the mobile chassis; The generation of the collaborator motion intention following speed command is based on the speed components of the end effector of the mobile manipulator in its base coordinate system. Define the speed direction of the projection of the end effector on the ground. The speed direction combines the compensation gain coefficient and the chassis yaw angle to generate the collaborator motion intention following compensation angular velocity. Through the compensation angular velocity and the lateral obstacle avoidance speed, obtain the obstacle avoidance speed required by the mobile chassis, that is, the collaborator motion intention following speed;

[0008] Step 3: Construct a kinematic model of the mobile manipulator, and design an augmented Jacobian null space control law in combination with the collaborator motion intention following speed to achieve autonomous obstacle avoidance of the mobile chassis on the premise of meeting the main human-robot collaboration task.

[0009] Further, in the step 1, establish a multi-level local environment perception model. Within the maximum detection range of the lidar of the mobile manipulator, the perception range of the model forms multiple sequentially increasing perception areas outward from the center point of the lidar to determine the urgency of obstacle avoidance.

[0010] Further, in the step 1, considering the actual obstacle avoidance requirements in the human-robot collaboration scenario, when the human-robot system approaches the wall, limited by the perception method of the single-line lidar, a long and continuous wall will reflect a large number of laser beams, resulting in the mobile chassis regarding the wall that can be driven close to as a huge obstacle and thus unable to approach. To solve this problem, use the hierarchical characteristics of the multi-level local perception model to design a sensitivity reduction area. Taking the center of the mobile chassis of the mobile manipulator as the coordinate origin, establish multiple sensitivity reduction areas. The sensitivity reduction area includes at least one outer layer of the perception area, thus ensuring that the obstacle perception in the first layer of the perception ring with the highest obstacle avoidance urgency is not affected.

[0011] Further, in the step 1, by collecting a one-dimensional array containing angle and distance information of the target, establish the relationship between the array index and the azimuth angle of the sampling point, and set the distance of the sampling point with an infinite return distance to the maximum detection distance.

[0012] Further, a circular range is generated with the installation center point of the mobile manipulator. Starting from the center point of the lidar of the mobile manipulator, tangents to the circle are connected to generate a shielding area. The lidar determines that there are no obstacles in this area, thereby removing the installation area of the manipulator and the working area guided by humans from the lidar sensing area.

[0013] Further, since this obstacle avoidance method is applicable to the human-machine collaborative operation scenario, the collaborator conducts collaborative guidance in the longitudinal direction (X-axis direction) of the omnidirectional mobile chassis, and the applicable mobile chassis is an omnidirectional mobile chassis with Mecanum wheels and has the ability to translate in any direction. Therefore, the local environment perception information provided by the lidar only needs to provide obstacle avoidance guidance in the lateral direction (Y-axis direction) of the chassis, that is, only the distance data on the Y-axis in the lidar coordinate system needs to be stored when storing lidar data.

[0014] Further, in step 2, a lateral obstacle avoidance speed command is generated by setting the obstacle avoidance weight values α, β, and γ from the inside to the outside, and:

[0015] α + β + γ = 1

[0016] 0 ≤ α ≤ β ≤ γ ≤ 1

[0017] The lateral obstacle avoidance linear speed of the mobile chassis is as follows:

[0018]

[0019]

[0020] Among them, Laserscan1*[i], Laserscan2*[i], and Laserscan3*[i] respectively represent the obstacle data storage arrays of the three-layer sensing rings, and G r is the desensitization coefficient, and m 1 , m 2 respectively represent the number of elements in the first and second rectangular desensitization regions DPC 1 [] and DPC 2 [] from the outside to the inside. v obs represents the lateral obstacle avoidance linear speed of the mobile chassis generated according to the multi-level local environment perception and the point cloud map of the rectangular desensitization region, which is weighted and synthesized by the obstacle avoidance speeds generated by each layer of sensing ring and the obstacle avoidance sensitive compensation speed generated by the rectangular desensitization region; v max represents the maximum allowable lateral movement speed of the mobile chassis; v y is the final lateral obstacle avoidance speed accepted by the mobile chassis.

[0021] Further, in step 2, a collaborator motion intention following speed command is generated by defining the speed of the end effector of the mobile manipulator in its base coordinate system as v ee, take its velocity components on the X-axis and Y-axis Define the velocity direction of the end effector projected on the ground:

[0022]

[0023] The collaborative motion intention following compensation angular velocity ω f is defined as follows:

[0024] ω f = k f ·(θ ee - θ ω )

[0025] where k f represents the compensation gain coefficient, and θ ω represents the chassis yaw angle;

[0026] To achieve autonomous obstacle avoidance of the mobile chassis during the human-robot collaborative handling process, the obstacle avoidance speed required for the mobile chassis is as follows:

[0027] v m = [0, v y , ω f T .

[0028] Furthermore, in step 3, the kinematic equation of the Mecanum wheel mobile manipulator is constructed as follows:

[0029]

[0030] where v e represents the velocity vector of the end of the mobile manipulator in the task space, q = [q m T , q A T T represents the generalized space coordinates of the mobile manipulator, and q A = [q 1 , q 2 , …, q n T respectively represent the joint space coordinates of the mobile platform and the manipulator, x m , y m represent the horizontal and vertical coordinates of the mobile platform, represents the rotation angle of the mobile platform, q 1 , q 2 , …, q n represent the joint angles of the manipulator, and n represents the number of joints of the manipulator, Denote the joint space velocity of the mobile manipulator, and J(q, φ) represents the Jacobian matrix of the mobile manipulator, and its elements satisfy a linear relationship with φ.

[0031] Furthermore, in step 3, design the augmented Jacobian null space control law as follows:

[0032]

[0033] where, Denote the joint velocity of the mobile manipulator of the final output, and respectively represent the cooperative task of the first priority and the obstacle avoidance task of the mobile chassis of the second priority, that is, the primary and secondary tasks. To ensure the convergence of the trajectory tracking errors of the end of the mobile manipulator and the mobile platform, and Denote the velocity commands of the end of the mobile manipulator and the mobile platform issued. The relationships between the above tasks and the joint velocities can be respectively expressed as:

[0034]

[0035] where, J 1 = J(q, φ), J 2 represents the Jacobian matrix of the secondary task. Since the constraints of the secondary task only act on the mobile chassis, its attitude is expressed as The Jacobian matrix J 2 of the mobile chassis obstacle avoidance task is as follows:

[0036] J 2 =(I 3×3 0 3×6 )

[0037] where, I 3×3 represents a 3×3 diagonal matrix, and 0 3×6 represents a 3×6 all-zero matrix;

[0038] In the human-machine cooperation scenario, by fusing the additional obstacle avoidance velocity v m of the mobile chassis and the mobile chassis velocity in the cooperative task joint velocity is obtained, and the formula is as follows:

[0039]

[0040] The advantages and beneficial effects of the present invention are as follows:

[0041] The autonomous obstacle avoidance method of a mobile manipulator based on null space control and intention guidance according to the present invention takes into account the actual obstacle avoidance situation in the human-machine collaboration scenario, can meet the obstacle avoidance requirements of the mobile chassis in the human-machine collaboration scenario, can remove the installation area of the manipulator and the working area guided by humans from the lidar sensing area, can reduce the amount of data stored in the lidar data, and finally realizes the autonomous obstacle avoidance of the mobile chassis during the human-machine collaborative handling process. Brief Description of the Drawings

[0042] Figure 1 It is a flowchart of the method according to an embodiment of the present invention.

[0043] Figure 2a It is a curve graph of the end trajectory tracking error in an embodiment of the present invention.

[0044] Figure 2b It is a curve graph of the mobile chassis trajectory tracking error in an embodiment of the present invention.

[0045] Figure 3 It is a broken line graph of the joint movement angle in an embodiment of the present invention.

[0046] Figure 4 It is a writing movement trajectory graph of the human-mobile manipulator in an embodiment of the present invention. Detailed Embodiment

[0047] The following will describe in detail the specific embodiments of the present invention with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for explaining and illustrating the present invention, and are not used to limit the present invention.

[0048] As Figure 1 shown, the autonomous obstacle avoidance method of a mobile manipulator based on null space control and intention guidance according to the present invention includes the following steps:

[0049] Step 1: Preprocess the single-line lidar point cloud data and construct a multi-level local environment perception point cloud map, and design an outer perception ring sensitivity reduction algorithm to meet the actual obstacle avoidance requirements in the human-machine collaboration scenario;

[0050] Step 1.1: The multi-level local environment perception point cloud map is constructed as follows:

[0051] The data returned by the single-line lidar is a one-dimensional array containing angle and distance information. Set the angle resolution to R θ = 360 / n, the number of sampling points n = 360, and the angle θ in the lidar coordinate system corresponding to the array index value i is θ = i·R θ . After establishing the relationship between the array index and the sampling point azimuth angle, preprocess the data returned by the lidar. When the lidar stores the detection data, it will store the data beyond the maximum detection distance dis maxThe return distance of the sampling points is set to Inf (infinity). To avoid the influence of the infinity value on the subsequent algorithm processing, the distance value of the sampling points with the return value of Inf is set to the maximum detection distance dis max .

[0052] After completing the preprocessing of the lidar data, a multi-level local environment perception model is established. The perception area of this model is divided into three levels. Within the maximum detection distance of the lidar, its detection range is divided into three circular areas with gradually increasing radii (R 1 < R 2 < R 3 ). A circular area is formed from the center of the lidar, and three perception rings with gradually decreasing gray levels are formed. The larger the gray level, the more sensitive it is to obstacles in this area, and the higher the urgency of obstacle avoidance. The red dots represent the distance values returned when the lidar laser beam is blocked by obstacles (objects composed of closed black solid lines) in this area. The detailed construction process of the multi-level local environment perception model is as follows: First, redefine the perception area of the lidar. Since it works in a human-robot collaboration scenario, the installation area of the robotic arm and the working area guided by humans need to be removed from the lidar perception area. The specific solution is to generate a circle with a radius of R arm with the center point of the robotic arm installation. Connect the tangent lines from the center point of the lidar to this circle to generate a shielding area with an angle of . The distance values detected by the lidar in this area are all returned as the maximum detection distance dis max , indicating that there are no obstacles in this direction.

[0053] After completing the redefinition of the perception area, convert the angle and distance data Laserscan[] in the polar coordinate system returned by the lidar into coordinates in the Cartesian coordinate system to obtain the X-axis coordinate matrix D x ∈ R 1×360 and the Y-axis coordinate matrix D y ∈ R 1×360 . Then, define the obstacle data storage arrays corresponding to the three layers of perception rings as Laserscan1[], Laserscan2[], and Laserscan3[] respectively. It should be noted that since this obstacle avoidance method is applicable to the human-robot collaborative operation scenario, the collaborator conducts collaborative guidance in the longitudinal direction (X-axis direction) of the omnidirectional mobile chassis, and the applicable mobile chassis is an omnidirectional mobile chassis with Mecanum wheels, which has the ability to translate in any direction. Therefore, the local environment perception information provided by the lidar only needs to provide obstacle avoidance guidance in the lateral direction (Y-axis direction) of the chassis, that is, only the distance data on the Y-axis in the lidar coordinate system needs to be stored when storing the lidar data. In summary, the pseudo-code of the point cloud data storage algorithm for each layer of the perception ring is as follows.

[0054]

[0055] After the point cloud data of each layer of the perception ring is stored, it is normalized to avoid the influence of its dimension on subsequent processing. The normalization method is shown in Equation (1):

[0056]

[0057] Step 1.2: Based on Step 1.1, design the outer layer perception ring sensitivity reduction algorithm as follows:

[0058] Considering the actual obstacle avoidance requirements in the human-machine collaboration scenario, when the human-machine system approaches the wall, due to the perception method of the single-line lidar, a long and continuous wall will reflect a large number of laser beams, causing the mobile chassis to regard the wall that can be driven close to as a huge obstacle and thus unable to approach. To solve this problem, a sensitivity reduction area is designed using the hierarchical characteristics of the multi-level local perception model. Taking the center of the mobile chassis as the origin of the coordinate system, two rectangular areas with lengths of l 1 and l 2 , and widths of h 1 and h 2 are established. These two rectangular areas only contain the second and third layers of the perception ring, thus ensuring that the obstacle perception in the first layer of the perception ring with the highest obstacle avoidance urgency is not affected. The pseudo-code of the outer layer perception ring sensitivity reduction algorithm is as follows:

[0059]

[0060] Step 2: Design the lateral obstacle avoidance speed command and the collaborator's motion intention following speed command to meet the obstacle avoidance requirements of the mobile chassis in the unstructured scenario;

[0061] Step 2.1: Design the lateral obstacle avoidance speed command as follows;

[0062] First, define the obstacle avoidance weight values of each layer of the perception ring as α, β, and γ in sequence. The relationship that the weight values should satisfy is shown in Equation (2), representing that the obstacle avoidance urgency increases in sequence from the inside to the outside.

[0063]

[0064] The designed lateral obstacle avoidance linear speed of the mobile chassis is shown in Equation (3), where G r is the sensitivity reduction coefficient, and m 1 and m 2 are the number of elements in the first and second layer rectangular sensitivity reduction areas DPC 1 [] and DPC 2 [] respectively. v obsThe lateral obstacle avoidance line speed of the mobile chassis generated based on multi-level local environment perception and the point cloud map of the rectangular desensitized area is synthesized by weighting the obstacle avoidance speeds generated by each layer of perception rings and the obstacle avoidance sensitivity compensation speed generated by the rectangular desensitized area; v max is the maximum lateral movement speed allowed for the mobile chassis; v y is the final lateral speed command received by the mobile chassis.

[0065]

[0066] Step 2.2: Design the collaborator's motion intention following speed command as follows:

[0067] Define the speed of the end effector of the mobile manipulator in its base coordinate system as v ee , and take its speed components on the X-axis and Y-axis Define the speed direction of the projection of the end effector on the ground as shown in Equation (4):

[0068]

[0069] The collaborator's motion intention following compensation angular velocity ω f is defined as shown in Equation (5), where k f is the compensation gain coefficient, and θ ω is the chassis yaw angle.

[0070] ω f = k f ·(θ ee - θ ω ) (5)

[0071] In summary, to achieve autonomous obstacle avoidance of the mobile chassis during the human-robot collaborative handling process, the obstacle avoidance speed required by the mobile chassis is as shown in Equation (6):

[0072] v m = [0, v y , ω f T (6)

[0073] Step 3: Construct the kinematic model of the mobile manipulator and design the augmented Jacobian null space control law to achieve autonomous obstacle avoidance of the mobile chassis while satisfying the main human-robot collaboration task.

[0074] Step 3.1: Construct the kinematic equation of the Mecanum wheel mobile manipulator as follows:

[0075]

[0076] where, v e represents the velocity vector of the end of the mobile manipulator in the task space, q = [qm T ,q A T T represents the generalized spatial coordinates of the mobile manipulator, and q A =[q 1 ,q 2 ,…,q n T respectively represent the joint space coordinates of the mobile platform and the manipulator, x m , y m represent the horizontal and vertical coordinates of the mobile platform, represents the rotation angle of the mobile platform, q represents the joint angle of the manipulator, n represents the number of joints of the manipulator, and in this patent, the research object n = 6, represents the joint space velocity of the mobile manipulator, J(q,φ) represents the Jacobian matrix of the mobile manipulator and its elements satisfy a linear relationship with φ.

[0077] Step 3.2: Design the augmented Jacobian null space control law as follows:

[0078]

[0079] In formula (8), is the finally output joint velocity of the mobile manipulator, and

[0080] respectively represent the cooperative task of the first priority and the mobile chassis obstacle avoidance task of the second priority (hereinafter referred to as the primary and secondary tasks). To ensure the convergence of the trajectory tracking errors of the end of the mobile manipulator and the mobile platform, and are the velocity commands issued for the end of the mobile manipulator and the mobile platform. The relationships between the above tasks and the joint velocities can be respectively expressed as shown in formula (9):

[0081]

[0082] where, J 1 = J(q,φ), J 2 is the Jacobian matrix of the secondary task. Since the constraints of the secondary task only act on the mobile chassis, its attitude can be expressed as Therefore, design the Jacobian matrix J 2 of the mobile chassis obstacle avoidance task as shown in formula (10):

[0083] J 2 =(I 3×3 0 3×6 ) (10)

[0084] ​​In the human-machine collaboration scenario applicable to the present invention, By fusing the additional obstacle avoidance speed v of the mobile chassis in formula (6) m and the mobile chassis speed in the collaborative task joint speed obtained, as shown in formula (11):

[0085]

[0086] Embodiment

[0087] The flow of the mobile manipulator autonomous obstacle avoidance method based on null space control and human intention guidance of the present invention is as Figure 1 shown. The specific object to be implemented is a mobile manipulator composed of a Mecanum wheel mobile platform and a six-degree-of-freedom manipulator, and its kinematic parameters are d b = 0.140m, d 1 = 0.140m, a 2 = 0.375m, a 3 = 0.345m, d 4 = 0.122m, d 5 = 0.122m, d 6 = 0.083m.

[0088] In the implementation process of the present invention, the desired trajectory of the collaborative task at the end of the mobile manipulator is designed as 0 ≤ t ≤ 20; the desired trajectory of the mobile chassis is designed as 0 ≤ t ≤ 20. Based on the above settings, the implementation effect of the augmented Jacobian null space control law of the mobile manipulator can be obtained. Figure 2a , Figure 2b illustrates that the augmented Jacobian null space control law of the mobile manipulator can accurately track the primary task trajectory and the secondary task trajectory; Figure 3 illustrates that the joint movement during the movement is relatively smooth; Figure 4 illustrates that the mobile manipulator autonomous obstacle avoidance method based on null space control and human intention guidance can enable the mobile manipulator to effectively avoid obstacles of the mobile chassis in an unstructured environment.

[0089] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. Autonomous obstacle avoidance method for mobile manipulator based on zero-space control and intention guidance, characterized by The steps include: Step 1: Obtain point cloud data and construct a multi-level local environment perception point cloud map. Based on the multi-layer perception range from the inside to the outside of the point cloud map, reduce the sensitivity of the outer perception range; Step 2: Generate lateral obstacle avoidance speed instructions and collaborator motion intention following speed instructions; the generation of lateral obstacle avoidance speed instructions is to define the obstacle avoidance weights of each layer of perception area to determine the urgency of obstacle avoidance from the inside to the outside, generate the lateral obstacle avoidance linear speed of the mobile chassis according to the multi-level local environment perception and the desensitization area point cloud map, and combine the maximum lateral movement speed allowed by the mobile chassis to obtain the final lateral obstacle avoidance speed accepted by the mobile chassis; the generation of the collaborator motion intention following speed instruction is to define the speed direction of the end effector projected on the ground by the speed component of the end effector of the mobile manipulator in its base coordinate system, and the speed direction is combined with the compensation gain coefficient and the chassis deflection angle to generate the collaborator motion intention following compensation angular velocity, and the obstacle avoidance speed required by the mobile chassis, that is, the collaborator motion intention following speed, is obtained through the compensation angular velocity and the lateral obstacle avoidance speed; Step 3: Construct a kinematic model of the mobile robot arm, and design an augmented Jacobi null space control law based on the collaborator's motion intention and speed following, so as to achieve autonomous obstacle avoidance of the mobile chassis while meeting the main task of human-machine collaboration.

2. The method for autonomous obstacle avoidance of a mobile robotic arm based on zero-space control and intention guidance according to claim 1, characterized in that: In the step 1, a multi-level local environment perception model is established. Within the maximum detection distance of the laser radar of the mobile robotic arm, the model perception range forms a plurality of successively increasing perception areas outward from the center point of the laser radar.

3. The method for autonomous obstacle avoidance of a mobile robotic arm based on zero-space control and intention guidance according to claim 2, characterized in that: In the step 1, the reduced sensitivity area is designed by utilizing the hierarchical characteristics of the multi-level local perception model, and a plurality of reduced sensitivity areas are established with the center of the mobile chassis of the mobile robot arm as the origin of the coordinate system. The reduced sensitivity area includes at least one outer perception area.

4. The method for autonomous obstacle avoidance of a mobile robotic arm based on zero-space control and intention guidance according to claim 1, characterized in that: In the step 1, a one-dimensional array containing angle and distance information of the target is collected, a relationship between the array index and the azimuth of the sampling point is established, and the distance of the sampling point with a returned distance of infinity is set as the maximum detection distance.

5. The method for autonomous obstacle avoidance of a mobile manipulator based on zero-space control and intention guidance according to claim 1, characterized in that: A circular range is generated with the center point of the mobile robot arm installation, and the tangent line of the circle is connected from the center point of the laser radar of the mobile robot arm to generate a shielding area. The laser radar determines that there are no obstacles in this area.

6. The method for autonomous obstacle avoidance of a mobile manipulator based on zero-space control and intention guidance according to claim 1, characterized in that: In the human-machine collaborative operation scenario, collaborators provide collaborative guidance in the longitudinal direction of the omnidirectional mobile chassis, and the applicable mobile chassis is an omnidirectional mobile chassis with Mecanum wheels, which has the ability to move horizontally in any direction. Therefore, the local environment perception information provided by the lidar only needs to provide obstacle avoidance guidance on the side of the chassis.

7. The method for autonomous obstacle avoidance of a mobile manipulator based on zero-space control and intention guidance according to claim 1, characterized in that: In step 2, a lateral obstacle avoidance speed instruction is generated by setting obstacle avoidance weights α, β and γ from inside to outside, and: α+β+γ=1 0≤α≤β≤γ≤1 The lateral obstacle avoidance linear speed of the mobile chassis is as follows: Among them, Laserscan1*[i], Laserscan2*[i], Laserscan3*[i] represent the obstacle data storage arrays of the three-layer perception ring, G r is the desensitization coefficient, m1 and m2 represent the number of elements in the first and second desensitization regions DPC1[] and DPC2[] from the outside to the inside, respectively, and v obs represents the lateral obstacle avoidance linear speed of the mobile chassis generated according to the multi-level local environment perception and the desensitization area point cloud map, which is a weighted synthesis of the obstacle avoidance speed generated by each layer of perception ring and the obstacle avoidance sensitive compensation speed generated by the rectangular desensitization area; v max Indicates the maximum lateral moving speed allowed by the mobile chassis; v y The final lateral obstacle avoidance speed accepted by the mobile chassis.

8. The method for autonomous obstacle avoidance of a mobile manipulator based on zero-space control and intention guidance according to claim 1, characterized in that: In step 2, the collaborator motion intention following speed instruction is generated by defining the speed of the end effector of the mobile robot arm in its base coordinate system as v ee , take its velocity components on the X-axis and Y-axis Define the velocity direction of the end effector projected on the ground: Collaborator motion intention follows compensation angular velocity ω f The definition is as follows: oh f =k f ·(θ ee -θ ω ) Among them, k f represents the compensation gain coefficient, θ ω Indicates chassis deflection angle; The obstacle avoidance speed required for the moving chassis is as follows: v m =[0,v y ,ω f ] T 。 9. The method for autonomous obstacle avoidance of a mobile robotic arm based on zero-space control and intention guidance according to claim 1, characterized in that: In step 3, the kinematic equation of the Mecanum wheeled mobile manipulator is constructed as follows: Among them, v e represents the velocity vector of the end of the mobile robot in the task space, q = [q m T ,q A T ] T represents the generalized space coordinates of the mobile robot, and q A =[q1,q2,…,q n ] T Represent the joint space coordinates of the mobile platform and the robotic arm, respectively, m ,y m Indicates the horizontal and vertical coordinates of the mobile platform, represents the rotation angle of the mobile platform, q1,q2,…,q n represents the joint angles of the robot arm, n represents the number of joints of the robot arm, represents the joint space velocity of the mobile robot arm, J(q,φ) represents the Jacobian matrix of the mobile robot arm and its elements satisfy the linear relationship with φ.

10. The method for autonomous obstacle avoidance of a mobile manipulator based on zero-space control and intention guidance according to claim 9, characterized in that: In step 3, the augmented Jacobian null space control law is designed as follows: in, Indicates the final output of the mobile robot joint velocity, and They represent the first-priority collaborative task and the second-priority mobile chassis obstacle avoidance task, i.e., the primary and secondary tasks, respectively. and It represents the speed command issued by the mobile robot end and the mobile platform. The relationship between the above tasks and joint speed can be expressed as: Among them, J1 = J(q, φ), J2 represents the Jacobian matrix of the secondary task, and its posture is expressed as The Jacobian matrix J2 of the mobile chassis obstacle avoidance task is as follows: J2=(I 3×3 0 3×6 ) Among them, I 3×3 represents a 3×3 diagonal matrix, 0 3×6 Represents a 3×6 all-0 matrix; In the human-machine collaboration scenario, By integrating the mobile chassis with the additional obstacle avoidance speed v m and the mobile chassis speed in the collaborative task joint speed The formula is as follows:

Images