A human-machine safety interactive control method

Abstract

The present invention relates to a human-machine safety interaction control method, including: obtaining proximity perception data and acceleration data; locating obstacles and obtaining static characteristic values ​​of obstacles; judging whether contact will occur; constructing a virtual repulsive field and defining the area where the virtual repulsive field is located as a dangerous operation area; regulating the speed within a safety threshold through a joint torque control method; detecting whether it touches an obstacle; identifying the touched obstacle and judging whether the obstacle is a person and the contact force exceeds the human-machine safety threshold. If so, emergency stop and return to the original route, otherwise, replan the path based on the virtual repulsive field. Compared with the prior art, the present invention proposes a fusion algorithm for multi-mode dynamic heterogeneous perception information of electronic skin, including virtual force field construction of multi-source heterogeneous information fusion, obstacle identity recognition, etc., which improves the utilization efficiency and perception accuracy of electronic skin information.

Description

Technical Field

[0001] The invention belongs to the technical field of robot motion control and relates to a human-machine safety interaction control method. Background Art

[0002] Safe and flexible operation is an important prerequisite for the widespread application of robots in the field of people's livelihood, among which human-machine safe interaction is an important indicator for the assessment of robot operation ability. This involves the robot's detection and positioning of relevant mobile personnel and obstacles in the working environment, as well as the robot's real-time obstacle avoidance in response to such situations.

[0003] At present, in the field of service robots, the robot obstacle avoidance method based on vision and radar sensors is widely used. Among them, vision and radar sensors have one or more problems such as being greatly affected by light, having perception blind spots, and high cost. The electronic skin system is an emerging and very promising robot perception system component that makes up for the shortcomings of vision and radar sensors. It can cover the entire body of the robot like human skin, realize the perception of touch, proximity, temperature and acceleration, and can generate a large amount of perception data in real time.

[0004] As a novel robot perception method, the electronic skin system is relatively lacking in means of information processing and utilization. Currently, most electronic skin models rely on a single modality perception source (proximity or touch) to achieve object detection, identification and motion feedback, which has the problems of low data utilization and low perception accuracy. Summary of the invention

[0005] The purpose of the present invention is to provide a human-machine safety interactive control method to overcome the problems of low data utilization and low perception accuracy.

[0006] The present invention uses a proximity sensor, an acceleration sensor, a pressure sensor and a temperature sensor in an electronic skin system to obtain proximity data, acceleration data, pressure data and temperature data.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] A human-machine safety interactive control method, the method comprising the following steps:

[0009] Step S1, obtaining proximity sense data and acceleration data;

[0010] Step S2: using proximity sensing data to locate the obstacle and obtain the static characteristic value of the obstacle;

[0011] Step S3: determine whether a collision will occur based on the vehicle's own motion trajectory and the static characteristic value of the obstacle. If so, proceed to step S4; if not, continue to move forward according to the preset trajectory.

[0012] Step S4, constructing a virtual repulsive field according to the proximity sense data and the acceleration data, and defining the area where the virtual repulsive field is located as a dangerous operation area;

[0013] Step S5: In the dangerous operation area, the speed is controlled within the safety threshold by means of the joint torque control method;

[0014] Step S6, obtaining pressure data and acceleration data, and detecting whether the vehicle has touched an obstacle based on the pressure data and acceleration data. If so, executing step S7; if not, continuing to move forward at the original speed;

[0015] Step S7, identify the touched obstacle to determine whether the obstacle is a person and whether the contact force exceeds the human-machine safety threshold. If so, execute step S8; otherwise, execute step S9;

[0016] Step S8, make an emergency stop and return to the original route;

[0017] Step S9, replan the path based on the virtual repulsive field, and return to step S1.

[0018] Furthermore, the step of obtaining the static characteristic value of the obstacle comprises the following steps:

[0019] Step S201, performing spatiotemporal registration on the proximity sense data to obtain registered proximity sense data;

[0020] Step S202: perform data fusion on the registered proximity data by using Kalman filtering and Gauss filtering methods, locate the obstacle, and obtain the center point of the obstacle;

[0021] Step S203: extracting static feature values ​​of the obstacle based on the center point position of the obstacle.

[0022] Furthermore, the static characteristic values ​​include characteristic values ​​of obstacle position, size and geometric shape.

[0023] Furthermore, the construction of the virtual repulsive field comprises the following steps:

[0024] Step S401, extracting proximity feature values ​​and acceleration feature values ​​from proximity sense data and acceleration data;

[0025] Step S402: Differentiate the approach characteristic value to obtain a velocity characteristic value relative to the obstacle;

[0026] Step S403: Using the mass-spring-damper model to obtain a virtual repulsive field, the approach eigenvalue, the velocity eigenvalue and the acceleration eigenvalue.

[0027] Furthermore, the detection of whether there is a contact with an obstacle includes using a hidden Markov model to fuse proximity data and acceleration data, making a probability judgment on the contact, and obtaining a detection result.

[0028] Furthermore, the identity identification comprises the following steps:

[0029] Step S701, performing spatiotemporal registration on the pressure data and the temperature data to obtain registered pressure data and registered temperature data;

[0030] Step S702, extracting pressure characteristic values ​​and temperature characteristic values ​​from the registered pressure data and the registered temperature data;

[0031] Step S703: Use the Bayes method and DS evidence theory to obtain the identity recognition result for the extracted pressure characteristic value and temperature characteristic value.

[0032] Furthermore, the re-planning of the path includes the following steps:

[0033] Step S901, defining a safe working area according to a virtual repulsive field;

[0034] Step S902: Use the objective function to redefine the ideal position in the safe working area.

[0035] In step S901 and step S902, the joint motion inequality is used to keep the joint angle of the vehicle within the physical space, and the joint torque control method is used to control the speed.

[0036] Furthermore, the objective function is:

[0037]

[0038] Among them, f i is the cost function, Z is the target expectation, w is the robot motion state parameter, x i It is an evaluation index function about the robot's motion state and force field.

[0039] Furthermore, the joint motion inequality is:

[0040] q min i ≤q i ≤q max i ,i=1,2,…,n

[0041] Among them, q min i is the minimum displacement of the i-th joint at the current moment, q max i is the maximum displacement of the i-th joint at the current moment.

[0042] Furthermore, the joint torque control method is as follows: mapping the force of the virtual repulsive field to the joint space to obtain the joint torque, and performing force control according to the joint torque. The calculation formula of the joint torque is:

[0043] τ ext =J s F ext

[0044] Among them, J s is the Jacobian matrix of the Cartesian joint space, F ext is the force of the virtual standpoint,

[0045] F ext =A·V

[0046] Among them, A is the force field parameter, V is the motion feature vector composed of acceleration data, velocity data and proximity data,

[0047]

[0048] A=<M,D,K>

[0049] in, q is acceleration data, speed data and proximity data, M, D and K are The coefficient of q.

[0050] Compared with the prior art, the present invention has the following characteristics:

[0051] 1. The present invention proposes a fusion algorithm for multi-modal dynamic heterogeneous perception information of electronic skin, including virtual force field construction and obstacle identification, which improves the utilization efficiency and perception accuracy of surrounding data information.

[0052] 2. The present invention conducts real-time full-range monitoring of obstacles around the robot and constructs a virtual repulsive field based on the obstacles. Based on this, the robot adjusts the running speed in real time to reduce the probability of emergencies. At the same time, when necessary, the robot performs secondary planning of the motion trajectory based on the virtual repulsive field to further improve the utilization efficiency and perception accuracy of the electronic skin information.

[0053] 3. The present invention uses pressure data and temperature data to simultaneously identify the touching object. During the contact process with the outside world, different contact force safety thresholds are set for people and objects respectively. When the contact force exceeds the threshold, the robot will take emergency measures to reduce the damage to a minimum, thereby improving the safety of the method. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1This is a schematic diagram of the flow chart of the human-machine safety interaction control method of the present invention;

[0055] Figure 2 This is a flow chart of a method for controlling speed by a joint torque control method of the present invention;

[0056] Figure 3 This is a flow chart of the joint torque control method of the present invention. DETAILED DESCRIPTION

[0057] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0058] Example:

[0059] The present invention provides a human-machine safety interaction control method, such as Figure 1 As shown, the human-machine safe interactive control method based on the environment-aware electronic skin system includes positioning obstacles in the robot's surrounding environment, building a virtual repulsive field based on the obstacles, adjusting the speed within the safety threshold in the dangerous area, re-planning the path based on the virtual repulsive field, emergency stop and return to the original path in an emergency, including the following steps:

[0060] Step S1, obtaining proximity sense data and acceleration data;

[0061] Step S2: using proximity sensing data to locate the obstacle and obtain the static characteristic value of the obstacle;

[0062] Step S3: determine whether a collision will occur based on the vehicle's own motion trajectory and the static characteristic value of the obstacle. If so, proceed to step S4; if not, continue to move forward according to the preset trajectory.

[0063] Step S4, constructing a virtual repulsive field according to the proximity sense data and the acceleration data, and defining the area where the virtual repulsive field is located as a dangerous operation area;

[0064] Step S5: In the dangerous operation area, the speed is controlled within the safety threshold by means of the joint torque control method;

[0065] Step S6, obtaining pressure data and acceleration data, and detecting whether the vehicle has touched an obstacle based on the pressure data and acceleration data. If so, executing step S7; if not, continuing to move forward at the original speed;

[0066] Step S7, identify the touched obstacle to determine whether the obstacle is a person and whether the contact force exceeds the human-machine safety threshold. If so, execute step S8; otherwise, execute step S9;

[0067] Step S8, make an emergency stop and return to the original route;

[0068] Step S9, replan the path based on the virtual repulsive field, and return to step S1.

[0069] Specifically, obtaining the static characteristic value of the obstacle comprises the following steps:

[0070] Step S201, performing spatiotemporal registration on the proximity sense data to obtain registered proximity sense data;

[0071] Step S202: perform data fusion on the registered proximity data by using Kalman filtering and Gauss filtering methods, locate the obstacle, and obtain the center point of the obstacle;

[0072] Step S203: extracting static feature values ​​of the obstacle based on the center point position of the obstacle.

[0073] More specifically, the static feature values ​​include feature values ​​of obstacle position, size and geometric shape.

[0074] Specifically, the construction of the virtual repulsive field comprises the following steps:

[0075] Step S401: extract the proximity characteristic value q and the corresponding acceleration characteristic value from the proximity sense data and the acceleration data , where q is the smallest approximate eigenvalue;

[0076] Step S402: Differentiate the approach characteristic value to obtain the velocity characteristic value relative to the obstacle. , acceleration characteristic value , velocity characteristic value and the close eigenvalue q constitute the motion eigenvector ;

[0077] Step S403: close characteristic value q, speed characteristic value and acceleration characteristic value Using the mass-spring-damper model, a virtual repulsive field is obtained.

[0078] Specifically, the control speed is within the safety threshold, see Figure 2 , including the following steps:

[0079] Step S501, calculate the position of the force in the virtual repulsive field and integrate the distributed force into the action force;

[0080] Step S502, determining the Jacobian matrix of each force in the Euclidean space mapped to the joint space;

[0081] Step S503: According to the force of the virtual repulsive field and the Jacobian matrix, joint force control is achieved through a joint torque control method, wherein when the proximity value between the robot and the obstacle is 0, the speed is also reduced to 0.

[0082] More specifically, see Figure 3 The joint torque control method is as follows: mapping the force of the virtual repulsive field to the joint space to obtain the joint torque, and performing force control according to the joint torque. The calculation formula of the joint torque is:

[0083] τ ext =J s F ext

[0084] Among them, J s is the Jacobian matrix of the Cartesian joint space, F ext is the force of the virtual standpoint,

[0085] F ext =A·V

[0086] Among them, A is the force field parameter, V is the motion feature vector composed of acceleration data, velocity data and proximity data,

[0087]

[0088] A=<M,D,K>

[0089] in, q is acceleration data, speed data and proximity data, M, D and K are The coefficient of q.

[0090] Specifically, the detection of whether the object touches an obstacle comprises the following steps:

[0091] Step S601: monitor the pressure data in the pressure sensor and the acceleration data in the acceleration sensor of the electronic skin system in real time.

[0092] Step S602: Use the hidden Markov model to fuse the proximity data and the acceleration data, make a probability judgment on the touch, and obtain a detection result.

[0093] Specifically, the identity identification includes the following steps:

[0094] Step S701, performing spatiotemporal registration on the pressure data and the temperature data to obtain registered pressure data and registered temperature data;

[0095] Step S702, extracting pressure characteristic values ​​and temperature characteristic values ​​from the registered pressure data and the registered temperature data;

[0096] Step S703: Use the Bayes method and DS evidence theory to obtain the identity recognition result for the extracted pressure characteristic value and temperature characteristic value.

[0097] Specifically, the re-planning of the path includes the following steps:

[0098] Step S901, defining a safe working area according to a virtual repulsive field;

[0099] Step S902: Use the objective function to redefine the ideal position in the safe working area.

[0100] In step S901 and step S902, the joint motion inequality is used to keep the joint angle of the vehicle within the physical space, and the joint torque control method is used to control the speed.

[0101] More specifically, the objective function is:

[0102]

[0103] Among them, f i is the cost function, Z is the target expectation, w is the robot motion state parameter, x i It is an evaluation index function about the robot's motion state and force field.

[0104] More specifically, the joint motion inequality is:

[0105] q min i ≤q i ≤q max i ,i=1,2,…,n

[0106] Among them, q min i is the minimum displacement of the i-th joint at the current moment, q max i is the maximum displacement of the i-th joint at the current moment.

[0107] In this embodiment, sensors in the electronic skin system are used to acquire multi-source data, and the sensors include a pressure sensor for collecting pressure data, a proximity sensor for collecting proximity data, a temperature sensor for collecting temperature data, and a six-axis acceleration sensor for collecting acceleration data.

[0108] The information collected by the above sensors includes: the relative position of the robot and the obstacle, the contact pressure when the robot contacts the obstacle, the temperature of the obstacle, and the six-axis acceleration of the robot during movement.

[0109] In a specific implementation, taking a robotic arm as an example, during the operation of the robot, the electronic skin generates a proximity event or a touch event and sends a data packet to the PC. The robot PC performs spatiotemporal registration of the acceleration data, proximity sense data, temperature data, and pressure data, and performs obstacle location, trajectory interference analysis, touch object identity recognition, and touch hazard assessment.

[0110] The robot and the obstacle position construct a virtual repulsive field, and the virtual repulsive field area is designated as a dangerous operation area. The robot converts the external force in the Cartesian space into the torque in the joint space, and adjusts the robot's movement speed to within the safety threshold; when the robot arm touches the object, the robot evaluates whether the touch force is within the safety threshold based on the identity of the touched object, and determines whether to perform emergency braking and return to the original route; when the trajectory interferes, the robot arm replans the movement trajectory based on the virtual repulsive field.

[0111] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program code.

[0112] The above description of the embodiments is to facilitate the understanding and use of the invention by those skilled in the art. It is obvious that those skilled in the art can easily make various modifications to these embodiments and apply the general principles described herein to other embodiments without creative work. Therefore, the present invention is not limited to the above embodiments, and improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope of the present invention should be within the scope of protection of the present invention.

Claims

1. A human-machine safety interactive control method, characterized in that: The method comprises the following steps: Step S1, obtaining proximity sense data and acceleration data; Step S2: using proximity sensing data to locate the obstacle and obtain the static characteristic value of the obstacle; Step S3: determine whether a collision will occur based on the vehicle's own motion trajectory and the static characteristic value of the obstacle. If so, execute step S4; if not, continue to move forward according to the preset trajectory. Step S4, constructing a virtual repulsive field according to the proximity sense data and the acceleration data, and defining the area where the virtual repulsive field is located as a dangerous operation area; Step S5: In the dangerous operation area, the speed is controlled within the safety threshold by means of the joint torque control method; Step S6, obtaining pressure data and acceleration data, and detecting whether the vehicle has touched an obstacle based on the pressure data and acceleration data. If so, executing step S7; if not, continuing to move forward at the original speed; Step S7, identify the touched obstacle to determine whether the obstacle is a person and whether the contact force exceeds the human-machine safety threshold. If so, execute step S8; otherwise, execute step S9; Step S8, make an emergency stop and return to the original route; Step S9, replanning the path based on the virtual repulsive field, and returning to step S1; The re-path planning includes the following steps: Step S901, defining a safe working area according to a virtual repulsive field; Step S902: Use the objective function to redefine the ideal position in the safe working area. In step S901 and step S902, the joint motion inequality is used to keep the joint angle of the robot in the physical space, and the joint torque control method is used to control the speed; The objective function is: Among them, f i is the cost function, Z is the target expectation, w is the robot motion state parameter, x i It is an evaluation index function about the robot's motion state and force field; The joint motion inequality is: q min i ≤q i ≤q max i ,i=1,2,…,n Among them, q min i is the minimum displacement of the i-th joint at the current moment, q max i is the maximum displacement of the i-th joint at the current moment; The joint torque control method is as follows: mapping the force of the virtual repulsive field to the joint space to obtain the joint torque, and performing force control according to the joint torque. The calculation formula of the joint torque is: t ext =J s F ext Among them, J s is the Jacobian matrix of the Cartesian joint space, F ext is the force of the virtual force field, F ext =A·V Among them, A is the force field parameter, V is the motion feature vector composed of acceleration data, velocity data and proximity data, A=<M,D,K> in, q is acceleration data, speed data and proximity data, M, D and K are The coefficient of q.

2. A human-machine safety interactive control method according to claim 1, characterized in that: The step of obtaining the static characteristic value of the obstacle comprises the following steps: Step S201, performing spatiotemporal registration on the proximity sense data to obtain registered proximity sense data; Step S202: perform data fusion on the registered proximity data by using Kalman filtering and Gauss filtering methods, locate the obstacle, and obtain the center point of the obstacle; Step S203: extracting static feature values ​​of the obstacle based on the center point position of the obstacle.

3. A human-machine safety interactive control method according to claim 2, characterized in that: The static characteristic values ​​include characteristic values ​​of obstacle position, size and geometric shape.

4. A human-machine safety interactive control method according to claim 1, characterized in that: The construction of the virtual repulsive field comprises the following steps: Step S401, extracting proximity feature values ​​and acceleration feature values ​​from proximity sense data and acceleration data; Step S402: Differentiate the approach characteristic value to obtain a velocity characteristic value relative to the obstacle; Step S403: Using the mass-spring-damper model to obtain a virtual repulsive field, the approach eigenvalue, the velocity eigenvalue and the acceleration eigenvalue.

5. The human-machine safety interactive control method according to claim 1, characterized in that: The detection of whether the object touches an obstacle includes using a hidden Markov model to fuse proximity data and acceleration data, making a probability judgment on the touch, and obtaining a detection result.

6. A human-machine safety interactive control method according to claim 1, characterized in that: The identification comprises the following steps: Step S701, performing spatiotemporal registration on the pressure data and the temperature data to obtain registered pressure data and registered temperature data; Step S702, extracting pressure characteristic values ​​and temperature characteristic values ​​from the registered pressure data and the registered temperature data; Step S703: Use the Bayes method and DS evidence theory to obtain the identity recognition result for the extracted pressure characteristic value and temperature characteristic value.

Images