CN110480678A - A kind of industrial robot collision checking method - Google Patents

A kind of industrial robot collision checking method Download PDF

Info

Publication number
CN110480678A
CN110480678A CN201910653305.5A CN201910653305A CN110480678A CN 110480678 A CN110480678 A CN 110480678A CN 201910653305 A CN201910653305 A CN 201910653305A CN 110480678 A CN110480678 A CN 110480678A
Authority
CN
China
Prior art keywords
torque
time
real
joint moment
practical
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201910653305.5A
Other languages
Chinese (zh)
Other versions
CN110480678B (en
Inventor
张冶
邵浩然
王杰高
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing Estun Robotics Co Ltd
Original Assignee
Nanjing Estun Robotics Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanjing Estun Robotics Co Ltd filed Critical Nanjing Estun Robotics Co Ltd
Priority to CN201910653305.5A priority Critical patent/CN110480678B/en
Publication of CN110480678A publication Critical patent/CN110480678A/en
Application granted granted Critical
Publication of CN110480678B publication Critical patent/CN110480678B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
    • B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
    • B25J19/00—Accessories fitted to manipulators, e.g. for monitoring, for viewing; Safety devices combined with or specially adapted for use in connection with manipulators
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
    • B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
    • B25J19/00—Accessories fitted to manipulators, e.g. for monitoring, for viewing; Safety devices combined with or specially adapted for use in connection with manipulators
    • B25J19/0095—Means or methods for testing manipulators

Landscapes

  • Engineering & Computer Science (AREA)
  • Robotics (AREA)
  • Mechanical Engineering (AREA)
  • Numerical Control (AREA)
  • Manipulator (AREA)

Abstract

The invention discloses a kind of industrial robot collision checking methods, using the collision checking method based on square difference, in robot operational process, predict the prediction torque in each joint in real time, and calculate the difference between prediction torque and practical denoising torque.Key point of the invention is the method with the correlation for calculating prediction torque and practical denoising torque, lag time of the practical denoising torque in each joint relative to prediction torque is precisely calculated, to do delay disposal to joint prediction torque, so that the two torque difference error is smaller, reduce collision threshold, improves the sensitivity of collision detection.

Description

A kind of industrial robot collision checking method
Technical field
It is especially a kind of for industrial robot collision detection the present invention relates to a kind of detection method of industrial robot Method.
Background technique
Industrial robot can collide with external object or people for some reason in operation process, and ontology is caused to damage Harmful and injury to personnel evil, therefore collision detection is particularly significant for the safety of robot.Common collision detection mainly can be with It is divided into two major classes, based on the collision detection of the sensors such as power feel, intelligent skin, detection sensitivity is can be improved in this mode, but It is to considerably increase cost；Another kind is detecting without sensor impact based on kinetic model, and this method can be reduced into This, but cause collision detection sensitivity not high due to error of modeling etc..
Chinese patent CN102426391A discloses one kind and is being run by calculating a collision scalar to measure robot In whether collide.Two variables, one is location error, and one is motor current value.And motor current value needs to increase Current sensor realizes that this will increase the cost of robot.
Chinese patent CN104985598B discloses a kind of industrial robot collision checking method, using based on torque difference Collision checking method, will after prediction torque postpones N number of period, calculate in real time joint ideal torque value and joint actual torque value it Between difference, when moment difference be more than collision threshold when, be considered as and collide.The invention will predict that joint moment postpones N backward A period treatment, but and it is undeclared delay period is calculated with what standard and method, be only to be obtained according to experimental analysis, ten Divide and be generally inaccurate, while robot, under the load of different operating condition difference postures, the response of servo-system is different therefore each The prediction torque delay time in joint is also not necessarily identical.Inaccurate delay disposal will cause the error of moment difference, increase Crash detection threshold, reduces the sensitivity of collision detection, is easy to make false alarm.
Summary of the invention
The technical problems to be solved by the invention are to overcome defect of the existing technology, provide a kind of suitable for appointing What robot without sensor, eliminate the high-precision collision checking method of delayed impact.
A kind of industrial robot collision checking method, includes the following steps:
Step 1, the practical joint moment M of robot is obtained by servo-drive in real time firstreal；
Step 2, practical joint moment when collecting in step 1 is filtered, removes influence of noise, obtains Practical denoising joint moment filt (Mreal)；
Step 3: calculating joint moment with Dynamic Models of Robot Manipulators, obtain joint prediction torque M；
Step 4: calculating prediction torque M relative to practical and denoise joint moment filt (Mreal) delay time td, root According to delay time tdWe correct prediction torque data M, make to predict that torque data M postpones t backwarddTime；
Step 5: by the prediction torque and reality denoising joint moment filt (M after delay disposalreal) make it is poor, be denoted as joint Torque difference △ M；
Step 6: setting the upper limit Thrd of collision thresholdupWith lower limit Thrddown；
Step 7: as △ M > ThrdupOr △ M < ThrddownWhen, it is considered as and collides, control system is sent to robot Stop signal, robot stop motion immediately.
Further, prediction torque M is calculated described in step 4 denoises joint moment filt relative to practical (Mreal) delay time td, include the following steps:
Step 4.1, prediction torque data M and practical denoising joint moment filt (M are calculatedreal) correlation ri, described Correlation riRefer in a period of time t1It is interior, it is located at prediction torque M and practical denoising joint moment filt on same time point (Mreal) difference quadratic sum；
The time t1It is greater than maximum possible delay time T, maximum delay time can be according to analysis of experimental results Or theoretical servo response and communication delay etc. are calculated；
Enable i=0；Minimum correlation value riFor Mmin；If Mmin=ri；
Step 4.2, make to predict that torque postpones certain time Δ t along the time axis；And i=i+1；
It calculates in time period t1It is interior, prediction torque along the time axis backward postpone Δ t duration after, prediction torque data M and Practical denoising joint moment filt (Mreal) correlation ri；
Step 4.3, the r that will be calculated in step 4.2iWith MminCompare；
Work as Mmin> riWhen, enable Mmin=ri, decay time=i, prediction torque delay time td=i* Δ t；
As i* Δ t < t1When, enter step 4.2；
As i* Δ t >=t1When, enter step 4.4；
Step 4.4, it exports: decay time=i, prediction torque delay time td=i* Δ t.
The utility model has the advantages that the present invention is based on the kinetic models of identification to calculate prediction torque, by detection prediction torque and For actual torque difference to determine whether colliding, method is simple and effective, without increasing any sensor, avoids increasing robot Structural complexity and raising production cost.
In addition the present invention considers the shadow that various factors lags system in view of the influence of control system feedback data lag Loud and Dynamic model error influence can accurately calculate each joint actual torque relative to prediction torque with high precision Lag time, and consider that actual torque influence of noise is filtered it, in this way it can be concluded that more accurate torque difference Value, improves the sensitivity of collision detection.
Key point of the invention is that each joint is precisely calculated with the method for the correlation for calculating two groups of data Lag time of the practical denoising torque relative to prediction torque, so that delay disposal is done to joint prediction torque, so that the two power Square mistake difference is smaller, reduces collision threshold, improves the sensitivity of collision detection.
Detailed description of the invention
Fig. 1 is industrial robot collision checking method schematic diagram of the present invention；
Fig. 2 is the step of present invention calculates prediction torque delay time figure；
Fig. 3 is the prediction torque and actual torque figure for not doing delay disposal；
Fig. 4 is the prediction torque and actual torque figure after doing delay disposal；
Fig. 5 is for the relational graph of the joint moment difference and collision threshold.
Specific embodiment
Fig. 1 is industrial robot collision checking method schematic diagram of the present invention, and the present invention is real with the first joint of industrial machine Object is tested, collision detection algorithm application of the invention is carried out.The following steps are included:
Step 1: practical joint moment M being obtained by servo-drive in real time firstreal。
Step 2: the practical joint moment acquired in step 1 being filtered, influence of noise is removed, is actually gone Make an uproar joint moment filt (Mreal)。
Step 3: calculating joint moment with Dynamic Models of Robot Manipulators, obtain joint prediction torque M.
Step 4: calculating delay time t of the prediction torque relative to practical denoising joint momentd, and to prediction torque its It is modified.
Fig. 2 is the specific steps figure for calculating prediction torque delay time, calculates prediction torque delay time tdThe step of such as Under:
Step 4.1, prediction torque data M and practical denoising joint moment filt (M are calculatedreal) correlation ri, described Correlation riRefer in a period of time t1It is interior, it is located at prediction torque M and practical denoising joint moment filt on same time point (Mreal) difference quadratic sum.
The time t1It is greater than maximum possible delay time T, maximum delay time can be according to analysis of experimental results Or theoretical servo response and communication delay etc. are calculated.
Enable i=0, minimum correlation value riFor Mmin, Mmin=ri；
Step 4.2, make to predict that torque postpones certain time Δ t along the time axis；And i=i+1；
That is, calculating in time period t1It is interior, after prediction torque postpones Δ t duration backward along the time axis, predict torque data M Joint moment filt (M is denoised with practicalreal) correlation ri；
Step 4.3, the r that will be calculated in step 4.2iWith MminCompare；
Work as Mmin> riWhen, enable Mmin=ri, decay time=i, prediction torque delay time td=i* Δ t；
As i* Δ t < t1When, enter step 4.2；
As i* Δ t >=t1When, enter step 4.4；
Step 4.4, export: decay time=i predicts torque delay time td=i* Δ t.
In this step 4, prediction torque data M and practical denoising joint moment filt (M are mainly calculatedreal) phase Pass value ri, the correlation riCalculation method can be using least square method, deviation integration method or other related coefficients etc. Method.It is described by taking deviation integration method as an example in the present embodiment, but is not limited to deviation integration method.
Predict torque data and practical denoising torque data filt (Mreal) correlation is stronger, then correlation riIt is smaller.First count Calculate a riThen value makes to predict that torque data M denoises joint moment filt (M to practicalreal) direction postpones with some cycles The Δ t time, such as it is 1ms that Δ t, which can be set, the correlation r after computing repeatedly the delay Δ t timei+1, compare two correlations Size, constantly denoise joint moment filt (M to practical by this methodreal) direction delay, delay total time is T time, is sought Find out correlation riThe smallest value Mmin, show that the number postponed at this time can calculate delay time td。
According to delay time tdWe correct prediction torque data, so that it is displaced t to actual torque directiondTime, in this way The influence that actual torque feedback lag generates can be eliminated, collision detection sensitivity is improved.
Fig. 3 is the prediction torque and actual torque for not doing delay disposal, and Fig. 4 is after acquiring delay time, at delay The prediction torque and actual torque comparison diagram of reason, it can be clearly seen that data coincidence is higher.
Step 5: by the prediction torque M and practical denoising torque filt (M after delay disposalreal) make it is poor.
Step 6: setting the upper limit Thrd of collision thresholdupWith lower limit Thrddown。
In the ideal case, in collisionless, torque difference is 0, then collision threshold may be configured as 0, but it is practical On due to identification model accuracy and other factors influence, prediction torque and practical denoising torque have a certain error, thus point Not She Zhi collision threshold upper limit ThrdupWith lower limit Thrddown。
Step 7: the torque difference of prediction torque and practical denoising torque after delay disposal is greater than ThrdupOr it is less than ThrddownWhen, it is considered as and collides, control system sends stop signal, robot stop motion immediately to robot.
Fig. 5 is the relational graph of the joint moment difference and collision threshold, when torque difference is more than that collision threshold bound is It can determine that position collides.

Claims (2)

1. a kind of industrial robot collision checking method, which comprises the steps of:
Step 1, practical joint moment M is obtained by servo-drive in real time firstreal；
Step 2, practical joint moment when collecting in step 1 is filtered, removes influence of noise, is actually gone Make an uproar joint moment filt (Mreal)；
Step 3: calculating joint moment with Dynamic Models of Robot Manipulators, obtain joint prediction torque M；
Step 4: calculating prediction torque M relative to practical and denoise joint moment filt (Mreal) delay time td, according to delay Time tdWe correct prediction torque data M, make to predict that torque data M postpones t backwarddTime；
Step 5: by the prediction torque and reality denoising joint moment filt (M after delay disposalreal) make it is poor, be denoted as joint moment Poor △ M；
Step 6: setting the upper limit Thrd of collision thresholdupWith lower limit Thrddown；
Step 7: as △ M > ThrdupOr △ M < ThrddownWhen, it is considered as and collides, control system is sent to robot is stopped Signal, robot stop motion immediately.
2. a kind of industrial robot collision checking method according to claim 1, which is characterized in that described in step 4 It calculates prediction torque M and denoises joint moment filt (M relative to practicalreal) delay time td, include the following steps:
Step 4.1, prediction torque data M and practical denoising joint moment filt (M are calculatedreal) correlation ri, the correlation Value riRefer in a period of time t1It is interior, it is located at prediction torque M and practical denoising joint moment filt (M on same time pointreal) The quadratic sum of difference；
The time t1It is greater than maximum possible delay time T, maximum delay time can be according to analysis of experimental results or theory Servo response and communication delay etc. are calculated；
Enable i=0；Minimum correlation value riFor Mmin；If Mmin=ri；
Step 4.2, make to predict that torque postpones certain time Δ t along the time axis；And i=i+1；
It calculates in time period t1It is interior, after prediction torque postpones Δ t duration backward along the time axis, predicts torque data M and actually go Make an uproar joint moment filt (Mreal) correlation ri
Step 4.3, the r that will be calculated in step 4.2iWith MminCompare；
Work as Mmin> riWhen, enable Mmin=ri, decay time=i, prediction torque delay time td=i* Δ t；
As i* Δ t < t1When, enter step 4.2；
As i* Δ t >=t1When, enter step 4.4；
Step 4.4, it exports: decay time=i, prediction torque delay time td=i* Δ t.
CN201910653305.5A 2019-07-19 2019-07-19 Industrial robot collision detection method Active CN110480678B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910653305.5A CN110480678B (en) 2019-07-19 2019-07-19 Industrial robot collision detection method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910653305.5A CN110480678B (en) 2019-07-19 2019-07-19 Industrial robot collision detection method

Publications (2)

Publication Number Publication Date
CN110480678A true CN110480678A (en) 2019-11-22
CN110480678B CN110480678B (en) 2022-03-04

Family

ID=68546158

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910653305.5A Active CN110480678B (en) 2019-07-19 2019-07-19 Industrial robot collision detection method

Country Status (1)

Country Link
CN (1) CN110480678B (en)

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111113488A (en) * 2019-12-30 2020-05-08 南京埃斯顿自动化股份有限公司 Robot collision detection device and method
CN112318501A (en) * 2020-10-23 2021-02-05 成都卡诺普自动化控制技术有限公司 Method for improving detection precision and protection sensitivity of collision force of robot
CN112936260A (en) * 2021-01-26 2021-06-11 华南理工大学 Sensor-free collision detection method and system for six-axis industrial robot
CN113628231A (en) * 2021-10-11 2021-11-09 中国人民解放军国防科技大学 Method and system for calculating impact center of small celestial body with unknown shape
WO2023123911A1 (en) * 2021-12-31 2023-07-06 达闼科技（北京）有限公司 Collision detection method and apparatus for robot, and electronic device and storage medium

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020045997A1 (en) * 1999-03-01 2002-04-18 Steven K. List Method and apparatus for adaptively learning test measurement delays on an individual device test for reducing total device test time
CN1849446A (en) * 2003-09-24 2006-10-18 株式会社爱安德 Multi-signal analysis device
CN101305177A (en) * 2005-11-08 2008-11-12 丰田自动车株式会社 Controller for internal combustion engine
CN101403924A (en) * 2008-11-03 2009-04-08 东华大学 Method for on-line delay detection of control loop
CN101793907A (en) * 2010-02-05 2010-08-04 浙江大学 Short-term wind speed forecasting method of wind farm
CN102753804A (en) * 2009-05-21 2012-10-24 丰田自动车株式会社 Device for estimating changes in target objects
CN104985598A (en) * 2015-06-24 2015-10-21 南京埃斯顿机器人工程有限公司 Industrial robot collision detection method
CN109361330A (en) * 2018-10-18 2019-02-19 东莞市旭展自动化科技有限公司 A kind of servo motor synchronisation control means based on bus

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020045997A1 (en) * 1999-03-01 2002-04-18 Steven K. List Method and apparatus for adaptively learning test measurement delays on an individual device test for reducing total device test time
CN1849446A (en) * 2003-09-24 2006-10-18 株式会社爱安德 Multi-signal analysis device
CN101305177A (en) * 2005-11-08 2008-11-12 丰田自动车株式会社 Controller for internal combustion engine
CN101403924A (en) * 2008-11-03 2009-04-08 东华大学 Method for on-line delay detection of control loop
CN102753804A (en) * 2009-05-21 2012-10-24 丰田自动车株式会社 Device for estimating changes in target objects
CN101793907A (en) * 2010-02-05 2010-08-04 浙江大学 Short-term wind speed forecasting method of wind farm
CN104985598A (en) * 2015-06-24 2015-10-21 南京埃斯顿机器人工程有限公司 Industrial robot collision detection method
CN109361330A (en) * 2018-10-18 2019-02-19 东莞市旭展自动化科技有限公司 A kind of servo motor synchronisation control means based on bus

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111113488A (en) * 2019-12-30 2020-05-08 南京埃斯顿自动化股份有限公司 Robot collision detection device and method
CN111113488B (en) * 2019-12-30 2022-06-14 南京埃斯顿自动化股份有限公司 Robot collision detection device and method
CN112318501A (en) * 2020-10-23 2021-02-05 成都卡诺普自动化控制技术有限公司 Method for improving detection precision and protection sensitivity of collision force of robot
CN112318501B (en) * 2020-10-23 2022-02-22 成都卡诺普机器人技术股份有限公司 Method for improving detection precision and protection sensitivity of collision force of robot
CN112936260A (en) * 2021-01-26 2021-06-11 华南理工大学 Sensor-free collision detection method and system for six-axis industrial robot
CN113628231A (en) * 2021-10-11 2021-11-09 中国人民解放军国防科技大学 Method and system for calculating impact center of small celestial body with unknown shape
WO2023123911A1 (en) * 2021-12-31 2023-07-06 达闼科技（北京）有限公司 Collision detection method and apparatus for robot, and electronic device and storage medium

Also Published As

Publication number Publication date
CN110480678B (en) 2022-03-04

Similar Documents

Publication Publication Date Title
CN110480678A (en) A kind of industrial robot collision checking method
US10324425B2 (en) Human collaborative robot system having improved external force detection accuracy by machine learning
US10839317B2 (en) Control device and machine learning device
Park et al. Learning-based real-time detection of robot collisions without joint torque sensors
CN108015774B (en) Sensor-free mechanical arm collision detection method
CN104985598A (en) Industrial robot collision detection method
Dimeas et al. Human-robot collision detection and identification based on fuzzy and time series modelling
CN109318232A (en) A kind of polynary sensory perceptual system of industrial robot
JP7072531B2 (en) Anomaly detection device and anomaly detection method
US20230115878A1 (en) Machine learning method for anomaly detection in an electrical system
CN113021340B (en) Robot control method, device, equipment and computer readable storage medium
CA2986207C (en) Failure diagnostic device and failure diagnostic method
CN112936260A (en) Sensor-free collision detection method and system for six-axis industrial robot
CN113365788A (en) Work discriminating apparatus and work discriminating method
CN114211491B (en) Collision detection method, device, computer readable storage medium and robot
WO2020094525A1 (en) Solution for machine learning system
CN117420792A (en) Central control system for composite processing machine tool
CN113984424B (en) Diagnostic method, device, equipment and storage medium for electronic cam function application
CN112123333A (en) Multi-joint robot collision detection method and device
CN109346200B (en) Nuclear power plant control rod static rod position linearity measuring method and electronic equipment
US11318606B2 (en) Controlling an automation assembly
Jung et al. Control of the manipulator position with the kinect sensor
CN115389240A (en) Robot joint detection method and device, computer equipment and readable storage medium
Ou et al. Online shock sensing for rotary machinery using encoder signal
Wu et al. An integral design for high performance sensor-less collision detection of serial robots

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant