CN109500811A - A method of the mankind are actively avoided towards man-machine co-melting robot - Google Patents
A method of the mankind are actively avoided towards man-machine co-melting robot Download PDFInfo
- Publication number
- CN109500811A CN109500811A CN201811342584.5A CN201811342584A CN109500811A CN 109500811 A CN109500811 A CN 109500811A CN 201811342584 A CN201811342584 A CN 201811342584A CN 109500811 A CN109500811 A CN 109500811A
- Authority
- CN
- China
- Prior art keywords
- robot
- route
- people
- mankind
- machine
- 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.)
- Pending
Links
Classifications
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1674—Programme controls characterised by safety, monitoring, diagnostic
- B25J9/1676—Avoiding collision or forbidden zones
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1656—Programme controls characterised by programming, planning systems for manipulators
- B25J9/1664—Programme controls characterised by programming, planning systems for manipulators characterised by motion, path, trajectory planning
- B25J9/1666—Avoiding collision or forbidden zones
Landscapes
- Engineering & Computer Science (AREA)
- Robotics (AREA)
- Mechanical Engineering (AREA)
- Manipulator (AREA)
Abstract
The invention proposes a kind of methods for actively avoiding the mankind towards man-machine co-melting robot.The position of this approach includes the following steps S1, the skeleton for being detected people using Kinect and calculating other side in real time, the position of other side's bone key point is estimated using improved particle filter algorithm, resettles manikin;The design manual parameter of S2, the bone key point of user and robot construct each section cylindrical body boundary, establish robot model;Whether will be collided between S3, detection cylinder body, then using the movement for analyzing people based on active collision-avoidance system, so that robot avoids;S4, it carries out planning new route in real time using dynamic route nomography, avoids robot actively.It is collided by detection cylinder body and is collided between people and robot to detect, so that robot can take appropriate action to avoid people.
Description
Technical field
The present invention relates to a kind of robot obstacle-avoiding technologies, and in particular to a kind of actively to avoid towards man-machine co-melting robot
The method of the mankind.
Background technique
With the development of industry 4.0, robot is in the following often intelligence and cooperation.Firstly, robot can be with people
Class interacts naturally.On the other hand, robot can cooperate with the personnel of public domain.From the perspective of the mankind, robot
For mankind's service, it means that the mankind and coexisting peacefully for robot are prerequisites existing for robot.In the future, robot and
The mankind would generally be used as affiliate's community life.The long term object of robot development is to realize to hand over the thought of people and spirit
Stream.Short-term goal is that robot can work together with the mankind, to complete the task different from the mankind.Therefore, man-machine collaboration
Concept is come into being, and gradually by industry, and academia and research institution pursue.Recent years, man-machine collaboration robot
It is mature, and used in some workshops, such as the production line that UR5 is Volkswagen.Man-machine collaboration will bring machine
Device people and robot are widely used
Robot is promoted to become partner indispensable in human lives.According to first principle of " three rule of robot ",
Robot may not injure the mankind or the observation mankind come to harm.Robot security is extremely important during industrial development.
It is well known that most of industrial robots have been placed into static, fixed " isolation " environment since industrial robot is born
In be more than half a century.In addition, they are always assigned a duplicate task to ensure their safety.Robot always by
Firm defence separates, this becomes the obstacle of advanced man-machine collaboration.In recent years, with the deep development of industry 4.0, manufacturing industry is opened
Begin to customize, personalized, flexibility constitutes severe challenge to traditional robot production method.In the development of automation,
All workers can not be replaced in a short period of time, to break economic law.Working portion is completed by robot.?
Before worker is replaced by robot completely, the interpersonal intimate contact of machine is inevitable.But traditional industry
Production method has been unable to meet the needs of enterprise's production safety.Safety have become an important factor for following man-machine interactive development it
One.Although man-machine collaboration allows robot to execute more extensively with more complicated task, it causes the safety problem of robot.
However, in the known integrated circuit it is a fact that increase the compromise that robot safety generally means that robot performance, this requires designer to find balance
Point to ensure safety and performance simultaneously.Now, cooperation robot can perceive environment according to the variation of environment and change them
Behavior.In addition, the development with industry 4.0 is replaced all existing in a short time since traditional robot is large number of
Traditional robot is clearly unpractical.Therefore, it is proposed to which a kind of effective method is possible between robot and the mankind to predict
The collision of generation, and provide feedback in time to ensure personal safety.For example, when robot arm hits human arm, machine
People can perceive the presence of human arm by force snesor.Then it stops or disappears to protect human security.This function limit
The performance of system cooperation robot, such as speed, load etc..When robot is unstructured, dynamically with working in human environment when,
Finding the mode that one kind safely and effectively works becomes more challenge.Safe and efficient human-computer interaction is for following robot
Using be it is indispensable, human body and robot must cooperate and could execute task jointly.It is wanted to meet most stringent of safety
It asks, robot must have the outstanding ability for changing path in complex environment.In anyone and robot cooperated system or people
In the environment coexisted with robot, remain that the safety of the mankind is a sizable challenge.In order to help the daily life of the mankind
It is living, had carried out some research and carried out development machines people's system, including with the mankind interact and cooperation.These research tables
Bright, the various tasks that robot can execute them together with the mankind are highly useful.Therefore, a kind of method is found to protect
Operator in human computer cooperation system is very important.This not only includes the possible collision of real-time detection, also avoids transporting
It collides when row with the mankind.
Real-time active anticollision is made of 3 parts: 1) environment sensing;2) it avoids collision;3) robot controls.Collision prevention is machine
One of most important field in people's technology.It is introduced in the first two work that there is many Planning concept of real-time capacity to be all based on
Potential situ method.In these methods, virtually attract and repel the field of force for indicating target and barrier.In such case
Under, the artificial target of machine attracts and far from barrier.
Real-time planning algorithm is critically important to robot system.Planning algorithm allows robot to avoid in true environment pair
As.A job description before is a kind of can adaptively the plan of landman and robot be calculated in hybrid precast system
Method.The advantages of another work report arm strength, can be used for detecting collision and avoiding obstacles.
Some researchs have evaluated the avoidance in dynamic environment.Human operator who is all considered as movement by the most people in them
Obstacle.This means that the collision that they only need to consider to avoid robot.But some researchs do not have.It is mentioned in a job
Method out analyzes human behavior and implements a model to simulate human behavior when colliding with robot.Another kind side
Method is that this distance is calculated by ambient image.Above-mentioned most of work assumes that the information in relation to environment is available.Collision avoids
Algorithm is based primarily upon the distance between the object in robot and true environment.
In human computer cooperation system, two kinds of methods avoided collision have been used for safety: the method for view-based access control model, analysis fortune
It is dynamic, color and texture, and the method based on inertial sensor, they use special motion capture.However, being passed based on inertia
The method of sensor be in the actual environment it is unpractical, because of the dedicated set that this method needs to have built-in sensors.Even if
In this way, it can only capture the movement of wearer rather than other his or her environment.This may cause serious problem, because all
The object enclosed may injure the mankind.Currently, the method for view-based access control model is the more preferable choosing avoided collision in robot control system
It selects.In addition, with the development of optics 3D sensor, some new depth transducers, such as Microsoft Kinect.
In recent years, the research of the method for view-based access control model concentrates on the efficiency for improving collision detection.A kind of method is related to one
Multi-camera system detects barrier.Another method is avoided collision using ad-hoc vision chip.Third uses three
Stereo visual system is weighed to track the movement for the operator being seated for wearing colour-coded.However, this triple stereoscopic visions
One the disadvantage is that monitored region may not show consistent color due to non-uniform ambient lighting conditions.This may
Mobile tracking is caused to fail.Another includes multiple 3D depth images.The third method is related to passing by using a Kinect
The integrated approach that sensor avoids collision, and also proposed deep space method in the work of same item.Although in safeguard protection
Aspect achieves many achievements, but most of real-time anticollision algorithms or too expensive or too limited, can not be in real world
It uses.These methods can not identify people or object, because they only detect mobile object.It is behaviour when robot takes for him or she
When making object, the mankind, which may find that, to be laid oneself open among danger.Due to lacking autonomous intelligence, avoid efficiency very low.
These methods can not identify people or object, because they only detect mobile object.It is operation pair when robot takes for him or she
As when, the mankind, which may find that, to be laid oneself open among danger.Due to lacking autonomous intelligence, avoid efficiency very low.These
Method can not identify people or object, because they only detect mobile object.When it is operation object that robot, which takes for him or she,
The mankind, which may find that, to be laid oneself open among danger.Due to lacking autonomous intelligence, avoid efficiency very low.
Summary of the invention
In order to solve the problems existing in the prior art described above, the present invention provides a kind of towards man-machine co-melting robot master
The method of the dynamic evacuation mankind, come the working space of supervisory-controlled robot and is detected possibly into machine using body-sensing sensor (Kinect)
Anyone of device people.When someone enters workspace, Kinect can detecte people and calculate the position of his or she bone in real time.
However, measurement error increases in quantity and amplitude over time due to tracking error and from the noise of equipment.
Therefore, the position of bone key point is estimated using improved particle filter algorithm.Can be used people bone key point and
The design manual parameter of robot constructs cylindrical body bounding box.People and machine are detected by the collision between detection cylinder body
It is collided between people, to improve the efficiency of collision detection.Then the movement of people is analyzed using rule-based flogic system,
So that robot can take appropriate action to avoid him or she, such as stop, in human peripheral's movement or far from people.Most
Afterwards, dynamic route nomography for planning new route to allow robot around remarkable in real time.According to active collision prevention, system can be true
People will not be met by protecting robot.
A method of the mankind actively being avoided towards man-machine co-melting robot, method includes the following steps:
The position of S1, the bone for detecting people using Kinect and calculating other side in real time, use unwise particle filter
(Unscented
Particle Filter, UPF) algorithm estimates the position of other side's bone key point, resettle manikin;
The design manual parameter of S2, the bone key point of user and robot constructs cylindrical body boundary, establishes machine
People's model;
Whether will be collided between S3, detection cylinder body, then analyze the movement of people using the rule-based decorum, obtain by
People's resistance
The route of gear, so that robot can take appropriate action to avoid him or she;
S4, it carries out planning new route to allow machine in real time using dynamic route nomography (Dynamic Roadmap)
People avoids robot actively by route around remarkable.
Further, enter the position of the skeleton of workspace described in step S1 using Kinect real-time detection, benefit
Human body bone is extracted in total with the interface that Kinect for Windows SDK 2.0 (Kinect Software Development Kit 2.0) is provided
15 joints of bone, be respectively as follows: head, shoulder center, right shoulder, right elbow, right finesse, left shoulder, left elbow, left finesse, buttocks center,
Right stern, right knee, right crus of diaphragm, left stern, left knee, left foot.The thickness of skeleton is estimated using unwise particle filter (UPF)
Meter finds one group of random sample propagated in state space by particle filter algorithm approximately to indicate probability density letter
Number replaces integral operation with sample average, and then obtains the minimum variance of system mode.Each joint of skeleton is made again
Manikin is established with cylinder, and the size of cylinder indicates the size in other side's difference joint, each cylinder connected by connecting rod.
Its is of different sizes in each joint, and if trunk is bigger than swivel of hand, cylinder is for simulating space shared by human body.
Further, step S2, which is established, obtains robot model, defines robot and machine according to controller convention first
Tool (mechanical arm) position of people and the direction of motion of tool.Using robot base as basis coordinates, basis coordinates to robot tool
The equation of motion of the continuous uniform transformation of coordinate can be defined as follows:
Wherein Ti i-1It is the translation matrix from i-1 coordinate to i coordinate, i indicates that i-th of link rod coordinate system, ν indicate robot
Speed, N is the quantity of joint (or joint coordinates).Following equation can be used to calculate in the position in j-th of joint:
Since the size in each joint can be obtained from the device parameter in the design manual that robot manufacturer provides, institute
Can determine the cylindrical radius in each joint.The length of boundary cylinder is the distance between the position of node j-1 and j.Then,
The cylindrical body in each joint can be constructed after the radius and length for determining cylinder, the cylindrical body for connecting each joint establishes machine
Device people's model.
Further, since human body and machine are modeled as the object being composed of cylindrical body per capita.Therefore step S3
It is middle to judge whether robot with human body has risk of collision, it is only necessary to which that whether detect between different cylindrical bodies has risk of collision;Cylinder
Whether the relative position between body can intersect according to cylinder and be divided into three classes, the bottom surface intersection of respectively two cylindrical bodies, two
A cylindrical body only intersects in a bottom surface of side intersection, cylindrical body with the side of another cylindrical body.
The rule of algorithm in the form of A so B usually by " IF A THEN B " i.e. if indicated, wherein A is leading item
Part, B are consequences.Algorithm mainly includes following several situations:
One, when the mankind are quickly i.e. with ν1Speed close to robot, the most wise selection of the situation is to avoid people;
Two, the mankind are slowly i.e. with ν2Speed it is close or people is collision prevention that is static, being made of unwise particle filter and Kinect
System will predict the motion profile of the mankind and use using expert system according to the current speed of the people tracked and position
Dynamic route nomography plans that a new route prevents from colliding with people, wherein ν1Greater than ν2。
Further, in continually changing environment, very big change can occur step S4 for angle and swing between joint
Change.The movement of robot in three dimensions is extremely complex.Since dynamic barrier may occur at any time, the fortune of robot
It is dynamic to be blocked, so, avoiding barrier is necessary to robot works in the environment of variation.The dynamic route
Nomography includes pretreatment stage and the planning of the adaptive real-time route of planning stage, the evacuation route for planning robot.
The algorithm improves the On-line efficiency of route planning using the precomputation of artificial potential field by generating pretreated route map;
Within given time, can with quickly determine route map before every wheel plan of robot which be partially prevented from, and by pre-
Processing stage and planning stage generate the dynamic map of enhancing.
Further, the pretreatment stage specific steps are as follows:
Step 1: the q dimension configuration space in robot constructs undirected road in the configuration space of robot working space
Line chart, wherein q is the quantity of joint of robot;
Step 2: searching the conflict point that may collide of machine person to person in route map and to look into the behavior found the right person pre-
The side on the route boundary of survey;Firstly, using the node of unified constant duration sampling method creation route map;Secondly, for every
A node identifies whether the column part between two nodes has collisionless possible and two sections according to pseudonorm space
The a part of the distance between point as route map, the node of this Some routes figure are also connected with k neighbor node;Defining A is machine
The set of all reference points on device people surface, a (p) are reference point a in the position of configuration space p, and a (q) is that reference point a is being configured
The position of space q, when robot is in configuration space q, following formula be in workspace two i.e. configuration space p and q of configuration it
Between robot deviate the maximum displacement of any reference point:
W represents the working space of the robot of robot;
After creating route map, the conflict point in route map and side, dynamic route nomography will be searched and deleted
(Dynamic Roadmap) is used to create the mapping in region between workspace grid and configuration space route map.
The planning stage is the following steps are included: delete the route stopped by people, by the current of the i.e. robot of target configuration
Configuration is connected to route map and searching for dynamic route figure avoids robot actively by route.
Compared with the prior art, the invention has the following advantages and beneficial effects:
1, the present invention prevents people and robot from colliding using Kinect real time monitoring, and it is any not need operator's wearing
Sensor, without invasive.
2, using the collision prevention method, it is only necessary to increase a Kinect sensor, configuration letter in the obstacle avoidance system of the invention
It is single at low cost.
3, a kind of robot people active anti-collision method proposed by the present invention.Since Kinect is a kind of very cheap and function
Powerful sensing system, therefore can be used for real-time detection human body.Then the behavior of people is analyzed, using one group of rule to be
System can be measured correctly.When the mankind are static, dynamic route figure (Dynamic Roadmap) algorithm is for direct
For the new path of robot planning.When the mankind slowly move, robot plans new route by using dynamic route nomography
To bypass it.When the mankind fast move, robot directly avoids the mankind immediately.By taking different measures, robot can
With more effective and safely avoid the mankind.Collision prevention is carried out since Kinect is only used only, at low cost and system configuration is simple.
Detailed description of the invention
Fig. 1 is robot danger zone schematic diagram in the present invention;
Fig. 2 is robot active collision avoidance system flow chart of the invention;
Fig. 3 is manikin schematic diagram of the invention;
Fig. 4 is mechanical arm schematic diagram of the invention;
Fig. 5 is the mechanical arm skeleton model figure after present invention modeling;
Fig. 6 is that cylinder of the invention intersects schematic diagram.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, with reference to the accompanying drawings and examples, right
Technical solution of the present invention is described in detail.
The invention proposes a kind of methods for actively avoiding the mankind towards man-machine co-melting robot.As shown in Figure 1, the party
Method is using body-sensing sensor (Kinect) come the working space of supervisory-controlled robot, when someone enters in robot workspace, people
In danger zone, Kinect can detecte people and calculate the position of his or she bone in real time.
As shown in Fig. 2, Kinect is a kind of very cheap and powerful sensing system, it can be used for catching in real time and arrest
The movement of body.The behavior of people is analyzed, using improved particle filter algorithm and algorithm so as to keeping away for planning robot
Touch route.When the mankind are static, dynamic route nomography (Dynamic Roadmap) is for being directly that robot planning is new
Path.When the mankind slowly move, using the behavior of expert system prediction people, the boundary of people is calculated according to the predictive behavior of people
Region, and robot is come by using Dynamic Roadmap planning new route around remarkable borderline region.When the mankind are quick
When mobile, robot directly avoids the mankind immediately.By taking different measures, robot more effectively and can be avoided safely
The mankind.Wherein, this method mainly includes following part:
S1, the position for detecting people using Kinect and calculating skeleton in real time, then use unwise particle filter
Algorithm estimates the position of bone key point;We enter the human body bone of workspace using Kinect real-time detection described in S1
Bone extracts 15 joints in total, is respectively as follows: first 1, shoulder center 2, right shoulder 3, right elbow 4, right finesse 5, left shoulder 6, left elbow 7, a left side
Wrist 8, buttocks center 9, right stern 10, right knee 11, right crus of diaphragm 12, left stern 13, left knee 14, left foot 15.Using improved unwise particle
Algorithm filter estimates bone thickness.Particle filter algorithm is propagated by resampling, is observed, and estimation, model is more
New step realizes the dynamically track of human body, be specifically exactly by find one group of random sample propagated in state space come
It is approximate to indicate probability density function, integral operation is replaced with sample average, and then obtain the minimum variance estimate of system mode
Process.Then each joint of bone is modeled using cylinder, modeling process is as follows: in every two adjacent joint
Point establishes a cylinder, and the size of cylinder connects depending on the size in each joint and by connecting rod by each cylinder,
If trunk is bigger than swivel of hand, for simulating space shared by human body, manikin as shown in Figure 3 is established with this.
The design manual parameter of S2, the bone key point of user and robot constructs cylindrical body boundary, establishes machine
People's model.Fig. 5 is the skeleton model figure after mechanical arm modeling, defines the position of robot and its tool according to controller convention first
It sets and its direction.Using robot base as basis coordinates, the continuous uniform of basis coordinates to robot tool coordinate is converted, the equation of motion
It can be defined as follows:
Wherein Ti i-1It is the translation matrix from i-1 coordinate to i coordinate, i indicates that i-th of link rod coordinate system, ν indicate robot
Speed, N is the quantity of joint (or joint coordinates).Following equation can be used to calculate in the position in j-th of joint:
Since the size in each joint can be obtained from the device parameter in the design manual that robot manufacturer provides, such as
Mechanical arm schematic diagram shown in Fig. 4, it is possible to determine the cylindrical radius in each joint.The length of boundary cylinder is node j-1
The distance between position of j.It is then possible to construct the cylindrical body in each joint after determining radius and length.
Whether will be collided between S3, detection cylinder body, then analyze the movement of people using rule-based system, so as to machine
Device people can take appropriate action to avoid human body;It is modeled as being composed of cylindrical body per capita due to human body and machine
Object.Therefore judge whether robot with people has risk of collision, it is only necessary to which whether detect between different cylindrical bodies has risk of collision.
Whether the cylindrical body of given two collisions, the relative position between them can intersect according to cylindrical seat and be divided into three classes, such as
Shown in Fig. 6, a respectively pedestal intersects with a pedestal, two cylindrical bodies are only in side intersection, one of cylindrical body
Base portion intersects these three situations with the side of another cylindrical body.
The rule of algorithm usually indicates that wherein A is antecedent condition in the form of " IF A THEN B ", and B is consequence.
Algorithm mainly includes following situation:
One, when the mankind are just in fast approaching robot, the most wise selection of the situation is to avoid people;Two, the mankind slowly connect
Closely, collision avoidance system will predict the motion profile of the mankind according to the current speed of the people tracked and position using expert system,
Borderline region is calculated, and plans that a new route prevents from colliding with people using dynamic route nomography;Three, the mankind are static
, a new route is planned using dynamic route nomography.
Two, S4, carry out planning new route to allow robot by route around remarkable in real time using dynamic route nomography;
In continually changing environment, great changes will take place for angle and swing meeting between joint.Robot is in three dimensions
It moves extremely complex.Since dynamic barrier may occur at any time, the movement of robot may be blocked.By this
Mode, avoiding barrier is necessary to robot works in the environment of variation.Using dynamic route nomography, one kind includes
Pretreatment stage and the planning of the adaptive real-time route of planning stage, are used for avoidance.The algorithm is by generating pretreated route
Figure calculates new path by dynamic route nomography according to pretreated route map using precomputation, to improve route
The On-line efficiency of planning.Before robot and people collide, which of route map can be quickly determined before every wheel plan
Partially it is prevented from a bit.The dynamic map of enhancing can be generated by pretreatment stage and planning stage.
In pretreatment stage, the first step is to tie up configuration space in the q of robot (to empty in matching for robot working space
Between) in undirected route map constructed by the relative position of robot and people and speed, wherein q is the quantity of joint of robot.The
Two steps are that the conflict point and the route for looking into the behavior prediction found the right person that machine person to person may collide are searched in route map
The side on boundary;Firstly, using the node of unified constant duration sampling method creation route map.Then, for each node, root
According to pseudonorm space, one that whether the part between two nodes has collisionless possible and this section as route map is identified
Divide k nearest-neighbors node of connection, carries out path planning.Define the set that A is all reference points on robotic surface, a (p)
It is the position of reference point a, a (q) is reference point a in the position of configuration space q.When robot is in configuration space q, following formula is
Any reference point may deviated utmostly between two configurations (i.e. configuration space p and q) in workspace.
If there is the boundary of a convex polyhedron in robot, and its all vertex is chosen for reference point, then the public affairs
Formula gives the maximum displacement of robot arbitrary point.Conflict point and side must be deleted when creating route map.Therefore, on creation road
After line chart, the conflict point in route map and side will be searched and delete, DRM is for creating workspace grid and configuration space route
The mapping in region between figure carries out active path planning.
Planning stage, by target configuration, (i.e. the current of robot was matched the following steps are included: deleting the route stopped by people
Set) be correspondingly connected with route map and search for dynamic route figure make robot by route actively avoid.
The above embodiment is a preferred embodiment of the present invention, but embodiments of the present invention are not by above-described embodiment
Limitation, other it is any without departing under Spirit Essence and principle of the invention it is made modification, modification, substitution, combination, simplify,
It should be equivalent substitute mode, should all be included within protection scope of the present invention.
Claims (6)
1. a kind of method for actively avoiding the mankind towards man-machine co-melting robot, which is characterized in that this method includes following step
It is rapid:
The position of S1, the skeleton for detecting people using Kinect and calculating other side in real time, use unwise particle filter
(Unscented Particle Filter, UPF) algorithm estimates the position of other side's bone key point, resettles manikin;
The design manual parameter of S2, the bone key point of user and robot construct each section cylindrical body boundary, establish machine
Device people's model;
Whether will be collided between S3, first detection cylinder body, then analyze the movement of people using rule-based system, obtained by people
The route of blocking, so that robot avoids;
S4, it carries out planning new route in real time using dynamic route nomography, avoids robot actively by route.
2. a kind of method for actively avoiding the mankind towards man-machine co-melting robot according to claim 1, feature exist
In entering the skeleton position of robot work region described in step S1 using Kinect real-time detection, utilize Kinect
The interface that for Windows SDK 2.0 is provided extracts 15 joints of skeleton in total, is respectively as follows: head (1), in shoulder
The heart (2), right shoulder (3), right elbow (4), right finesse (5), left shoulder (6), left elbow (7), left finesse (8), buttocks center (9), right stern
(10), right knee (11), right crus of diaphragm (12), left stern (13), left knee (14), left foot (15);Using unwise particle filter (UPF) to people
The thickness of body bone estimated, by particle filter algorithm find one group propagated in the state space of skeleton with
Press proof indicates probability density function originally, replaces integral operation with sample average, and then obtain the variance of state, realize human body with
Track;Manikin is established into each joint of skeleton using cylinder again, the size of cylinder indicates other side's difference joint
Size is connected each cylinder by connecting rod.
3. a kind of method for actively avoiding the mankind towards man-machine co-melting robot according to claim 1, feature exist
In step S2's establishes robot model, defines each connecting rod of robot according to the controller convention of robot itself first
The tool location of coordinate system and robot, and the direction of motion of tool;Using robot base as basis coordinates, basis coordinates to robot
The equation of motion of tool coordinates continuous uniform transformation is as follows:
Wherein Ti i-1It is the translation matrix from i-1 coordinate to i coordinate, i indicates that i-th of link rod coordinate system, ν indicate the speed of robot
Degree, N is the quantity of joint or joint coordinates, and the position of j-th of articulation nodes is calculated with following equation:
The size in each joint of robot is obtained from the device parameter in the design manual that robot manufacturer provides, and is thereby determined that
The cylindrical radius in each joint in robot model;The length of boundary cylinder is the distance between the position of node j-1 and j;?
Determine cylinder radius and length after construct the cylindrical body in each joint, the cylindrical body for connecting each joint establishes robot mould
Type.
4. a kind of method for actively avoiding the mankind towards man-machine co-melting robot according to claim 1, feature exist
In detecting whether the mode that will be collided in step S3 are as follows: whether intersect the relative position between two cylindrical bodies;Intersection is divided into
Three classes, the bottom surface intersections of respectively two cylindrical bodies, two cylindrical bodies only side intersection, cylindrical body a bottom surface with it is another
Intersect the side of a cylindrical body.
If algorithm is indicated in the form of A so B, wherein A is antecedent condition, and B is consequence, and algorithm mainly includes such as
Lower situation:
One, when people is with ν1Speed close to robot, then robot avoids people;Two, when people is with ν2Speed close to robot
Or people be it is static, what the collision avoidance system being made of unwise particle filter and Kinect will be current according to the people tracked
Speed and position are predicted the motion profile of the mankind using expert system and plan a new route using dynamic route nomography
It prevents from colliding with people, wherein ν1Greater than ν2。
5. a kind of method for actively avoiding the mankind towards man-machine co-melting robot according to claim 1, feature exist
In dynamic route nomography described in step S4 includes pretreatment stage and planning stage, the evacuation road for planning robot
Line;Dynamic route nomography generates pretreated route map by pretreatment stage, is mentioned using the precomputation of artificial potential field
The On-line efficiency of high route planning;The route being prevented from route map is determined before every wheel plan evacuation of robot, and is led to
It crosses pretreatment stage and the planning stage generates dynamic map.
6. a kind of method for actively avoiding the mankind towards man-machine co-melting robot according to claim 1 or 5, feature
It is, the pretreatment stage specific steps are as follows:
Step 1: the q dimension configuration space in robot constructs undirected route map in the configuration space of robot working space,
Wherein q is the quantity of joint of robot;
Step 2: searching the conflict point and look into the behavior prediction found the right person that machine person to person may collide in route map
The side on route boundary;Firstly, using the node of unified constant duration sampling method creation route map;Secondly, for each section
Point, according to pseudonorm space, identify the column part between two nodes whether have it is collisionless may and two nodes it
Between a part as route map of distance, the node of this Some routes figure is also connected with k neighbor node;Defining A is robot
The set of all reference points on surface, a (p) are reference point a in the position of configuration space p, and a (q) is reference point a in configuration space
The position of q, when robot is in configuration space q, following formula is to configure machine i.e. between configuration space p and q at two in workspace
Device people deviates the maximum displacement of any reference point:
W represents the working space of robot;
After creating route map, the conflict point in route map and side, dynamic route nomography (Dynamic will be searched and deleted
Roadmap) for creating the mapping in region between workspace grid and configuration space route map.
The planning stage is the following steps are included: delete the route stopped by people, by the target configuration i.e. current-configuration of robot
It is connected to route map and searching for dynamic route figure avoids robot actively by route.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811342584.5A CN109500811A (en) | 2018-11-13 | 2018-11-13 | A method of the mankind are actively avoided towards man-machine co-melting robot |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811342584.5A CN109500811A (en) | 2018-11-13 | 2018-11-13 | A method of the mankind are actively avoided towards man-machine co-melting robot |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109500811A true CN109500811A (en) | 2019-03-22 |
Family
ID=65748068
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811342584.5A Pending CN109500811A (en) | 2018-11-13 | 2018-11-13 | A method of the mankind are actively avoided towards man-machine co-melting robot |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109500811A (en) |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110253570A (en) * | 2019-05-27 | 2019-09-20 | 浙江工业大学 | The industrial machinery arm man-machine safety system of view-based access control model |
CN110561432A (en) * | 2019-08-30 | 2019-12-13 | 广东省智能制造研究所 | safety cooperation method and device based on man-machine co-fusion |
CN112140102A (en) * | 2020-06-08 | 2020-12-29 | 深圳市越疆科技有限公司 | Obstacle avoidance method, device and system of industrial robot |
CN112817302A (en) * | 2019-11-18 | 2021-05-18 | 中冶赛迪工程技术股份有限公司 | Safety control method, system, equipment and readable medium for industrial robot |
CN112883792A (en) * | 2021-01-19 | 2021-06-01 | 武汉海默机器人有限公司 | Robot active safety protection method and system based on visual depth analysis |
CN114451991A (en) * | 2021-01-30 | 2022-05-10 | 威海威高骨科手术机器人有限公司 | Fracture reduction path planning method based on virtual reduction collision detection |
CN114872043A (en) * | 2022-05-09 | 2022-08-09 | 苏州艾利特机器人有限公司 | Robot collision detection method, storage medium and electronic device |
JP2023033033A (en) * | 2021-08-26 | 2023-03-09 | 正▲うえ▼精密工業股▲ふん▼有限公司 | Obstacle avoidance method of robot arm and obstacle avoidance system of robot arm |
CN115890671A (en) * | 2022-11-17 | 2023-04-04 | 山东大学 | SMPL parameter-based multi-geometry human body collision model generation method and system |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105137973A (en) * | 2015-08-21 | 2015-12-09 | 华南理工大学 | Method for robot to intelligently avoid human under man-machine cooperation scene |
CN106125925A (en) * | 2016-06-20 | 2016-11-16 | 华南理工大学 | Method is arrested based on gesture and voice-operated intelligence |
CN107206592A (en) * | 2015-01-26 | 2017-09-26 | 杜克大学 | Special purpose robot's motion planning hardware and production and preparation method thereof |
CN107544514A (en) * | 2017-09-29 | 2018-01-05 | 广州唯品会研究院有限公司 | Robot barrier thing preventing collision method, device, storage medium and robot |
CN108527370A (en) * | 2018-04-16 | 2018-09-14 | 北京卫星环境工程研究所 | The man-machine co-melting safety control system of view-based access control model |
-
2018
- 2018-11-13 CN CN201811342584.5A patent/CN109500811A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107206592A (en) * | 2015-01-26 | 2017-09-26 | 杜克大学 | Special purpose robot's motion planning hardware and production and preparation method thereof |
CN105137973A (en) * | 2015-08-21 | 2015-12-09 | 华南理工大学 | Method for robot to intelligently avoid human under man-machine cooperation scene |
CN106125925A (en) * | 2016-06-20 | 2016-11-16 | 华南理工大学 | Method is arrested based on gesture and voice-operated intelligence |
CN107544514A (en) * | 2017-09-29 | 2018-01-05 | 广州唯品会研究院有限公司 | Robot barrier thing preventing collision method, device, storage medium and robot |
CN108527370A (en) * | 2018-04-16 | 2018-09-14 | 北京卫星环境工程研究所 | The man-machine co-melting safety control system of view-based access control model |
Non-Patent Citations (1)
Title |
---|
张苗辉: "基于视觉系统的行人检测与跟踪方法研究", 《中国博士学位论文全文数据库信息科技辑》 * |
Cited By (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110253570A (en) * | 2019-05-27 | 2019-09-20 | 浙江工业大学 | The industrial machinery arm man-machine safety system of view-based access control model |
CN110561432A (en) * | 2019-08-30 | 2019-12-13 | 广东省智能制造研究所 | safety cooperation method and device based on man-machine co-fusion |
CN112817302A (en) * | 2019-11-18 | 2021-05-18 | 中冶赛迪工程技术股份有限公司 | Safety control method, system, equipment and readable medium for industrial robot |
CN112140102A (en) * | 2020-06-08 | 2020-12-29 | 深圳市越疆科技有限公司 | Obstacle avoidance method, device and system of industrial robot |
CN112883792A (en) * | 2021-01-19 | 2021-06-01 | 武汉海默机器人有限公司 | Robot active safety protection method and system based on visual depth analysis |
CN114451991A (en) * | 2021-01-30 | 2022-05-10 | 威海威高骨科手术机器人有限公司 | Fracture reduction path planning method based on virtual reduction collision detection |
CN114451991B (en) * | 2021-01-30 | 2024-02-27 | 威海威高骨科手术机器人有限公司 | Fracture reduction path planning method based on virtual reduction collision detection |
JP7307776B2 (en) | 2021-08-26 | 2023-07-12 | 正▲うえ▼精密工業股▲ふん▼有限公司 | Obstacle avoidance method for robot arm and obstacle avoidance system for robot arm |
JP2023033033A (en) * | 2021-08-26 | 2023-03-09 | 正▲うえ▼精密工業股▲ふん▼有限公司 | Obstacle avoidance method of robot arm and obstacle avoidance system of robot arm |
WO2023217032A1 (en) * | 2022-05-09 | 2023-11-16 | 苏州艾利特机器人有限公司 | Robot collision detection method, storage medium and electronic device |
CN114872043B (en) * | 2022-05-09 | 2023-11-17 | 苏州艾利特机器人有限公司 | Robot collision detection method, storage medium and electronic equipment |
CN114872043A (en) * | 2022-05-09 | 2022-08-09 | 苏州艾利特机器人有限公司 | Robot collision detection method, storage medium and electronic device |
CN115890671A (en) * | 2022-11-17 | 2023-04-04 | 山东大学 | SMPL parameter-based multi-geometry human body collision model generation method and system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109500811A (en) | A method of the mankind are actively avoided towards man-machine co-melting robot | |
Han et al. | Dynamic obstacle avoidance for manipulators using distance calculation and discrete detection | |
US8154590B2 (en) | Method for observation of a person in an industrial environment | |
Schmidt et al. | Depth camera based collision avoidance via active robot control | |
EP2952993B1 (en) | Method for building a map of probability of one of absence and presence of obstacles for an autonomous robot | |
Pyo et al. | Service robot system with an informationally structured environment | |
CN101479082B (en) | Robot device and robot device control method | |
Krüger et al. | Image based 3D surveillance for flexible man-robot-cooperation | |
KR20200054327A (en) | 3D space monitoring device, 3D space monitoring method, and 3D space monitoring program | |
Blodow et al. | Perception and probabilistic anchoring for dynamic world state logging | |
Hermann et al. | Anticipate your surroundings: Predictive collision detection between dynamic obstacles and planned robot trajectories on the GPU | |
Rezazadegan et al. | Risked-based design for the physical human-robot interaction (pHRI): An overview | |
Tang et al. | The integration of contactless static pose recognition and dynamic hand motion tracking control system for industrial human and robot collaboration | |
Lemmerz et al. | A hybrid collaborative operation for human-robot interaction supported by machine learning | |
Manns et al. | Identifying human intention during assembly operations using wearable motion capturing systems including eye focus | |
KR20220133810A (en) | Utility Vehicle and Corresponding Apparatus, Method and Computer Program for a Utility Vehicle | |
Suchan et al. | Commonsense scene semantics for cognitive robotics: Towards grounding embodied visuo-locomotive interactions | |
Ding et al. | Monocular camera-based complex obstacle avoidance via efficient deep reinforcement learning | |
Teke et al. | Real-time and robust collaborative robot motion control with Microsoft Kinect® v2 | |
Deeken et al. | SEMAP-a semantic environment mapping framework | |
Lee et al. | Collision detection system for the practical use of the humanoid robot | |
Singh et al. | Understanding socially aware robot navigation | |
Chikhalikar et al. | An object-oriented navigation strategy for service robots leveraging semantic information | |
Dylla et al. | A qualitative representation of social conventions for application in robotics | |
Heindl et al. | Metric pose estimation for human-machine interaction using monocular vision |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190322 |