WO2019010612A1

WO2019010612A1 - Robot joint anti-collision protection system and method based on sensing fusion technology

Info

Publication number: WO2019010612A1
Application number: PCT/CN2017/092395
Authority: WO
Inventors: 程时胜; 王驰; 许晨举; 唐良成; 张帝
Original assignee: 深圳市艾唯尔科技有限公司
Priority date: 2017-07-10
Filing date: 2017-07-10
Publication date: 2019-01-17
Also published as: CN107466263A

Abstract

The invention discloses a robot joint anti-collision protection system and method based on the sensing fusion technology, comprising sensors arranged on a plurality of joints of the robot respectively, wherein an output end of the sensor is sequentially connected with a data fusion module, a feature extraction module, a plurality of pattern recognition modules, an information fusion module, an information extraction module and a decision-making module. Hierarchical multi-sensor information fusion target tracking is adopted to achieve information acquisition of the system, and a prediction residual error is introduced to modify the tracking residual error, which cannot reflect the error caused by an unpredictable information completely, so the prediction residual error is adopted to modify a correction term. With the system and method, the obstacle information can be acquired in real time and effectively, and a reliable information source is provided for the safety decision.

Description

Robot joint anti-collision protection system and method thereof for sensor fusion technology

[0001] The present invention relates to the field of robot control technologies, and in particular, to a robot joint anti-collision protection system and a method thereof.

Background technique

[0002] During normal operation of the robot, if there is external interference, the robot control system may be disordered, resulting in the robot not working properly. The robot components are often collided, causing operator injury or damage to the robot.

technical problem

[0003] Robot anti-collision system came into being, but the design of the existing robot anti-collision system lacks unified and comprehensive planning, resulting in the same information target, the information is incomplete, the information is inconsistent, and even The problem of information. The robot collision avoidance system lacks comprehensive and accurate grip on the information target, and has poor ability to respond to abnormal events.

Problem solution

Technical solution

[0004] In view of the above-mentioned deficiencies of the prior art, the technical problem to be solved by the present invention is to provide a robot joint anti-collision protection system and a method thereof for sensing fusion technology, and collecting surrounding data by using various sensors provided on the robot. The parameters are used to remind the operator or adopt the method of the robot's own power-off protection to realize the self-protection of the robot and protect the operator's beneficial effects to solve the deficiencies of the prior art.

[0005] In order to achieve the above object, the present invention provides a robot joint anti-collision protection system with sensor fusion technology, characterized in that a sensor fusion technology robot joint anti-collision protection system is characterized in that a sensor disposed on a plurality of joints of the robot, wherein the sensor output end is connected to the data fusion module, and the data fusion module performs overall analysis on the data collected by the plurality of sensors, and the data fusion module is configured with multiple stages of comparison parameters. And comparing the collected data with the comparison parameter; the output end of the data fusion module is connected to the feature extraction module, and is compared with the data collected by the sensor by the data fusion module, and then transmitted to the feature extraction module, and the feature extraction module Output and multiple modes The identification module is connected, and the feature extraction module outputs according to the unused data result according to the specific pattern recognition module; the output of the plurality of pattern recognition modules is connected to the information fusion module, and the output end of the information fusion module is connected with the information extraction module. The output of the information extraction module is connected to the decision module; the sensor includes but is not limited to: a pyroelectric sensor, an infrared proximity sensor, an ultrasonic sensor, an image sensor, an acceleration sensor, and a speed sensor, wherein the pyroelectric sensor is used to detect whether There are obstacles, the image sensor is used to detect the position of the obstacle, the acceleration sensor, the speed sensor is used to detect the acceleration and speed of the robot joint, the infrared proximity sensor or the ultrasonic sensor is used to detect the distance between the robot and the obstacle. These data are made for the system control robot. Provide reference data for the obstacle avoidance response.

[0006] The robot joint anti-collision protection system of the above-mentioned sensor fusion technology is characterized in that the sensor is disposed on different joint regions of the robot, and the joint region includes but is not limited to an upper joint, an arm joint, a wrist joint, The data fusion module summarizes the sensor data at different locations, and the single set of sensor acquisition data has the same contrast parameters as the data fusion module. The feature extraction module (outputs according to different pattern recognition modules according to the settings).

[0007] The robot joint anti-collision protection system of the above-mentioned sensor fusion technology is characterized in that the decision module includes but is not limited to a voice alarm module and a servo motion driving module.

[0008] A method for protecting a joint of a robot joint by a sensor fusion technology, comprising the following steps

[0009] Step one, sensor layer tracking, in the sensor tracking, the data acquired by the local sensor is processed by the node to obtain a local state estimation;

[0010] Step 2: information fusion tracking, the results of each tracking and state estimation of the sensor layer are sent to the information fusion module, and a global state estimation is obtained through fusion, and the result is outputted on the one hand and feedback to the sensor layer on the other hand. Node

[0011] Step 3. Perform a risk estimation based on the global state estimation, and make an alarm decision if there is a dangerous situation.

[0012] The robot joint anti-collision protection method of the above-mentioned sensor fusion technology is characterized in that the data acquired by the local sensor of the first step is ambient environment dynamic information and robot state parameters, wherein the surrounding environment dynamic information includes whether There are obstacle information, obstacle position information, and robot state parameters include robot acceleration and speed information, robot and obstacle distance information.

[0013] The robot joint anti-collision protection method of the above sensor fusion technology is characterized in that: The global state estimation of the second is equivalent to the sum of the fusion results of the state estimation and the correction. The correction includes two parts: one is the sum of the local sensor tracking and the tracking residual error; the other is the global prediction and the local node prediction residual error. sum.

[0014] The above-described sensor joint fusion technology robot joint anti-collision protection method is characterized in that: in the third step, there is a dangerous situation to make an alarm decision, specifically: the robot estimates the parameter in the global state to reach a dangerous threshold, and issues Voice prompt alarm to the staff.

[0015] The above-described sensor joint fusion technology robot joint anti-collision protection method is characterized in that, in the third step, if there is a dangerous situation to make an alarm decision, the robot also controls the joint with danger of collision. Avoid actions.

Advantageous effects of the invention

Beneficial effect

[0016] The present invention adopts hierarchical multi-sensor information fusion target tracking to realize system information acquisition, and introduces prediction residual error to modify the tracking residual error, which cannot completely reflect the error caused by unpredictable information, and uses the prediction residual error to modify the correction. Item, has a clearer physical meaning, and is easier to implement. Obtain obstacle information in a practical and effective manner, providing a reliable source of information for security decisions.

The concept, the specific structure, and the technical effects produced by the present invention will be further described in conjunction with the accompanying drawings in order to fully understand the objects, features and effects of the invention.

Brief description of the drawing

DRAWINGS

1 is a system block diagram of the present invention.

2 is a flow chart of an implementation method of the present invention.

3 is an effect diagram of the implementation of the present invention.

4 is a schematic view showing the distribution of the sensor of the present invention.

Embodiments of the invention

[0022] As shown in FIG. 1 to FIG. 4, a robot joint anti-collision protection system of a sensing fusion technology is characterized in that a robot joint anti-collision protection system of a sensing fusion technology is characterized in that: In The sensor 1 is connected to the data fusion module 2, and the data fusion module 2 performs overall analysis on the data collected by the plurality of sensors 1. The data fusion module 2 has multiple settings. The comparison parameter of the phase is compared with the comparison parameter by collecting the data; the output end of the data fusion module 2 is connected with the feature extraction module 3, and is compared with the data collected by the sensor by the data fusion module 2, and then transmitted to the feature extraction. Module 3, the output end of the feature extraction module 3 is connected to a plurality of pattern recognition modules 4, and the feature extraction module 3 outputs according to a specific pattern recognition module 4 according to the unused data result; the output ends of the plurality of pattern recognition modules are The information fusion module 5 is connected, the output of the information fusion module 5 is connected to the information extraction module 6, and the output of the information extraction module 6 is connected to the decision module 7; the sensor 1 includes but is not limited to: pyroelectric sensor, infrared Proximity sensor, ultrasonic sensor, image sensor, acceleration sensor, speed sensor, wherein The discharge sensor is used to detect whether there is an obstacle, the image sensor is used to detect the position of the obstacle, the acceleration sensor, the speed sensor is used to detect the acceleration and speed of the robot joint, the infrared proximity sensor or the ultrasonic sensor is used to detect the distance between the robot and the obstacle, These data provide reference data for the system to control the robot to make obstacle avoidance reactions.

[0023] In this embodiment, the sensor 1 is disposed on different joint regions on the robot, and the joint region includes but is not limited to an upper joint, an arm joint, and a wrist joint, and the data fusion module 2 summarizes sensor data at different positions. The single set of sensor acquisition data is the same as the comparison parameter inside the data fusion module 2, and the feature extraction module 3 performs output by using different pattern recognition modules 4 according to the settings. Sensors placed in these joint areas can detect obstacles around the robot, obstacles, obstacles, obstacles and robot distances, motion acceleration and speed of the robot joints.

[0024] In this embodiment, the decision module 7 includes, but is not limited to, a voice alarm module and a servo motion driving module. Among them, the voice alarm module realizes the voice dialogue function between the robot and the human, and the steering motion driving drive module realizes the robot anti-collision avoidance function.

[0025] As shown in FIG. 2, a robot joint anti-collision protection method for sensing fusion technology is characterized in that it comprises the following steps:

[0026] Step one, sensor layer tracking, in the sensor tracking, the data acquired by the local sensor is processed through the node to obtain a local state estimation;

[0027] Step 2: information fusion tracking, the results of each tracking and state estimation of the sensor layer are sent to the information fusion module, and a global state estimation is obtained through fusion, and the result is on the one hand as an output result, and the other side The surface will be fed back to each node of the sensor layer;

[0028] Step 3. Perform a risk estimation based on the global state estimate, and make an alarm decision if there is a dangerous situation.

[0029] In this embodiment, the data acquired by the local sensor of the first step is ambient environment dynamic information and a robot state parameter, wherein the surrounding environment dynamic information includes whether there is an obstacle information, an obstacle position information, and the robot state parameter includes Robot acceleration and speed information, robot and obstacle distance f π.

[0030] In this embodiment, the global state estimation of the second step is equivalent to the sum of the fusion results of the state estimation and the correction, wherein the correction includes two parts: one is the sum of the local sensor tracking and the tracking residual error; The second is the sum of global prediction and local node prediction residual error.

[0031] In this embodiment, the alarming decision is made in the third step in the dangerous situation: the robot obtains a voice prompt alarm to the staff after the global state estimation parameter reaches the dangerous threshold.

[0032] In this embodiment, if there is a dangerous situation in the third step to make an alarm decision, the robot also controls the joint with danger of collision to make an avoidance action.

[0033] As shown in FIG. 4, the robot body of the present invention is provided with a plurality of sensors, wherein the pyroelectric sensor is used for detecting whether there is an obstacle, the image sensor is used for detecting an obstacle position, and the acceleration sensor and the speed sensor are used for Detecting the acceleration and speed of the robot joint, the infrared proximity sensor or the ultrasonic sensor is used to detect the distance between the robot and the obstacle. These data provide reference data for the system to control the robot to avoid obstacles.

[0034] The data collected by the sensor is sent to the data fusion module of the CPU, the feature extraction module, the plurality of pattern recognition modules, the information fusion module, and the information extraction module to perform processing in sequence. The use of hierarchical multi-sensor information fusion, target tracking to achieve system information acquisition, and the introduction of predictive residual error to modify tracking residual error can not fully reflect the error caused by unpredictable information. Among them, the tracking residual error reflects the error caused by the unpredictable information in the global tracking step, and thus is used to correct the global state estimation. However, since local state estimation is related to local and global tracking, tracking residual error cannot fully reflect the error caused by unpredictable information. Therefore, using the prediction residual error to modify the correction term has a clearer physical meaning and is easier to implement. Obtain obstacle information in a practical and effective manner and provide a reliable source of information for security decisions.

[0035] As shown in FIG. 3, the voice system of the robot can communicate with customers in various ways, and reminds the user to note. The intention to collide with the robot, to remind the user and communicate with the user. The robot is equipped with the same sensor. The control system also sets the safety distance to detect the object close to the robot. When the distance parameter estimated by the robot in the global state reaches the dangerous threshold, a voice prompt alarm is given to the staff, such as "You want "I met me" to remind the user to make a evasive action. If a more dangerous situation occurs, such as when the robot has been slightly bumped, the robot will send out various exchanges such as "My hand is broken" and so on. In this case, the robot is turned off by controlling the steering gear at the joint where the collision is dangerous, and the avoidance action is described above. The preferred embodiment of the present invention has been described in detail above. It will be appreciated that many modifications and variations can be made by those skilled in the art without departing from the scope of the invention. Therefore, any technical solution that can be obtained by a person skilled in the art according to the prior art by logic analysis, reasoning or limited experiment according to the prior art should be within the scope of protection determined by the claims.

Claims

Claim

[Claim 1] 1. A robot joint anti-collision protection system for sensing fusion technology, comprising: a sensor (1) respectively disposed on a plurality of joints of a robot, the sensor (1) output end and data Fusion module (2) connection, data fusion module (2) The overall data of the data collected by multiple sensors 1 is analyzed. The data fusion module (2) is set with multiple stages of comparison parameters, by collecting data and comparing parameters. Performing an alignment; the output of the data fusion module (2) is connected to the feature extraction module (3), and is compared with the data collected by the sensor by the data fusion module (2), and then transmitted to the feature extraction module (3). The output end of the feature extraction module (3) is connected to a plurality of pattern recognition modules (4), and the feature extraction module ( ₃ ) outputs according to a specific pattern recognition module (4) according to the unused data result; The output of the identification module is connected to the information fusion module (5), and the output of the information fusion module (5) is connected to the information extraction module (6), and the information extraction module ( 6) The output is connected to the decision module (7); the sensor (1) includes but is not limited to: a pyroelectric sensor, an infrared proximity sensor, an ultrasonic sensor, an image sensor, an acceleration sensor, a speed sensor, wherein the pyroelectric sensor For detecting the presence of obstacles, an image sensor is used to detect the position of the obstacle, an acceleration sensor, a speed sensor is used to detect the acceleration and speed of the joint of the robot, an infrared proximity sensor or an ultrasonic sensor is used to detect the distance between the robot and the obstacle, and the data is a system. The control robot provides reference data for obstacle avoidance response.

2. A robot joint anti-collision protection system with sensor fusion technology according to claim

The sensor is disposed on different joint regions of the robot, and the joint region includes but is not limited to an upper joint, an arm joint, and a wrist joint, and the data fusion module aggregates sensor data at different positions, and the single one is The collected data of the group sensor is the same as the comparison parameter inside the data fusion module, and the feature extraction module outputs according to the different pattern recognition modules set.

3. The robot joint anti-collision protection system of the sensor fusion technology according to claim 1, wherein the decision module comprises, but is not limited to, a voice alarm module and a servo motion driving module.

4. A robot joint anti-collision protection method for sensing fusion technology, characterized in that it comprises the following steps:

Step 1: Tracking of the sensor layer. In the sensor tracking, the data acquired by the local sensor is processed through the node to obtain a real local state estimation;

Step 2: Information fusion tracking, the results of each tracking and state estimation of the sensor layer are sent to the information fusion module, and a global state estimation is obtained through fusion, and the result is outputted on the one hand, and is fed back to each node of the sensor layer;

Step 3. Make a risk estimate based on the global state estimate and make an alarm decision if there is a dangerous situation.

The robot joint anti-collision protection method of the sensor fusion technology according to claim 4, wherein the data acquired by the local sensor of the first step is ambient environment dynamic information and robot state parameters, wherein the surrounding environment The dynamic information includes information on whether there is an obstacle, obstacle position information, and robot state parameters include robot acceleration and speed information, robot and obstacle distance information.

The robot joint anti-collision protection method of the sensor fusion technology according to claim 4, wherein the global state estimation of the second step is equivalent to the sum of the fusion results of the state estimation and the correction, wherein The correction consists of two parts: one is the sum of local sensor tracking and tracking residual error; the other is the sum of global prediction and local node prediction residual error.

The robot joint anti-collision protection method of the sensor fusion technology according to claim 4, wherein the step 3 has a dangerous situation to make an alarm decision, specifically: the robot estimates the parameter in a global state to reach danger Threshold 吋, a voice prompt alarm is given to the staff.

The robot joint anti-collision protection method of the sensor fusion technology according to claim 4, wherein in the third step, if there is a dangerous situation to make an alarm decision, the robot also controls the collision risk. The joints make a evasive action.