CN105467926A - Motion control method and device and artificial intelligence device - Google Patents
Motion control method and device and artificial intelligence device Download PDFInfo
- Publication number
- CN105467926A CN105467926A CN201410459517.7A CN201410459517A CN105467926A CN 105467926 A CN105467926 A CN 105467926A CN 201410459517 A CN201410459517 A CN 201410459517A CN 105467926 A CN105467926 A CN 105467926A
- Authority
- CN
- China
- Prior art keywords
- information
- motion
- target travel
- movable information
- movable
- 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
Links
Landscapes
- Manipulator (AREA)
- Toys (AREA)
Abstract
The invention provides a motion control method and device and an artificial intelligence device. The method comprises that different motion information corresponding to different sample individuals collected by a sensor is obtained; target motion information used to optimizing corresponding motion is determined according to the different motion information; and the target motion information is sent to the artificial intelligence device, so that the artificial intelligence device implements the corresponding motion according to the target motion information. According to the invention, the optimization efficiency of the artificial intelligence device is improved.
Description
Technical field
The present invention relates to smart machine technology, particularly a kind of motion control method and device, artificial intelligence equipment.
Background technology
Along with the development of science and technology, currently the mankind can be replaced to perform some tasks by artificial intelligence equipment such as robot, such as robot can cook, or robot also can wash clothes, or robot also can do some in the factory concerning the dangerous action of tool the mankind, to avoid the danger of the mankind.Therefore, robot is very helpful to human lives or work tool.
But, in current technology, the movement control mode of robot is relatively simple, such as in robot, embed fixing executive routine, correlated condition in this image data and executive routine, after the exogenous data receiving sensor collection, compares by the controller of robot interior, is obtained the corresponding action performed by simple Logic judgment, such as, and the execution of this action is the action parameter according to being stored in advance in robot, walks according to the walking stride value stored.But robot differs according to the action that this parameter performs and obtains good movement effects surely, such as, because the improper task of making robot perform of walking stride value can not well complete.Now can optimize its action by the internal processes and corresponding action parameter revising robot, the motion of robot self can be improved and improve, but obviously, this mode needs to carry out modification of program and debugging, likely amendment repeatedly and debugging just can reach a good movement effects, very complicated, makes the action optimization efficiency of robot very low, and the business that affects performs.
Summary of the invention
In view of this, the invention provides a kind of motion control method and device, artificial intelligence equipment, to improve the optimization efficiency of artificial intelligence equipment.
Particularly, the present invention is achieved through the following technical solutions:
First aspect, provides a kind of motion control method, comprising:
Obtain multiple movable informations that sensor gathers, the plurality of movable information is corresponding multiple individual of sample respectively;
According to described multiple movable information, determine the target travel information optimizing corresponding motion;
Described target travel information is sent to artificial intelligence equipment, with the motion making described artificial intelligence equipment corresponding according to described target travel information and executing.
Second aspect, provides a kind of motion control method, comprising:
Obtain target travel information, described target travel information is that motion control device obtains according to multiple movable informations of sensor collection, described multiple movable information corresponding multiple individual of sample respectively;
The motion corresponding according to described target travel information and executing, with by motion corresponding to multiple movable information described in described target travel Advance data quality.
The third aspect, provides a kind of motion control device, comprising:
Information receiving module, for multiple movable informations that receiving sensor gathers, described multiple movable information corresponding multiple individual of sample respectively;
Message processing module, for according to described multiple movable information, determines the target travel information optimizing corresponding motion;
Information sending module, for described target travel information is sent to artificial intelligence equipment, with the motion making described artificial intelligence equipment corresponding according to described target travel information and executing.
Fourth aspect, provides a kind of artificial intelligence equipment, comprising:
Data obtaining module, for obtaining target travel information, described target travel information is that motion control device obtains according to multiple movable informations of sensor collection, described multiple movable information corresponding multiple individual of sample respectively;
Motor execution module, for the motion corresponding according to described target travel information and executing, with by motion corresponding to multiple movable information described in described target travel Advance data quality.
The motion control method that the embodiment of the present invention provides and device, artificial intelligence equipment, it is the movable information of the multiple individual of samples according to sensor collection, be optimized this target travel information of moving, owing to there being the foundation optimized, this target travel information can be obtained faster, and also more meet the motion of individual of sample, relative to the continuous amendment debugging of individual equipment of the prior art, improve the optimization efficiency of artificial intelligence equipment.
Accompanying drawing explanation
Fig. 1 is the application scenarios one of the motion control method that the embodiment of the present invention provides;
Fig. 2 is the process flow diagram of a kind of motion control method that the embodiment of the present invention provides;
Fig. 3 is the process flow diagram of the another kind of motion control method that the embodiment of the present invention provides;
Fig. 4 is the application scenarios two of the motion control method that the embodiment of the present invention provides;
Fig. 5 is the data coupling schematic diagram in the motion control method that provides of the embodiment of the present invention;
Fig. 6 is the structural representation of the motion control device that the embodiment of the present invention provides;
Fig. 7 is the structural representation of the artificial intelligence equipment that the embodiment of the present invention provides;
Fig. 8 is the entity structure schematic diagram of the motion control device that the embodiment of the present invention provides;
Fig. 9 is the entity structure schematic diagram of the artificial intelligence equipment that the embodiment of the present invention provides.
Embodiment
The parameters such as artificial intelligence equipment such as robot, when performing certain action, being need according to some action datas, such as, supposing that its arm will lift by robot, the angle that the height needing foundation arm to lift, arm lift; Suppose that robot will walk, need according to parameters such as the step pitch of walking, the speed of walking.The parameters such as above-mentioned height, speed are stored in robot interior, is such as stored in for the memory module in the controller of control motion.
But in existing routine techniques, the above-mentioned parameter for control motion is relatively-stationary, once by this optimum configurations in robot, robot just needs to move according to this parameter, and amendment parameter is also more loaded down with trivial details.The movement control mode of the embodiment of the present application, will obtain the above-mentioned action parameter for optimizing robot motion exactly as soon as possible, robot is improved fast and perfect.It should be noted that, in each embodiment of follow-up the application, artificial intelligence equipment is described with the artificial example of machine, actual implement in be not limited to this, also can be the similar artificial intelligence equipment of other control principles.
The motion control method of the embodiment of the present application, by gathering the movable information of a large amount of individual of samples, and the comprehensive controling parameters obtaining robot motion according to these information fast.Concrete, see Fig. 1, show the optional application scenarios of one of the motion control method of the present embodiment, first this scene under simple declaration.
Choose the individual of sample for gathering movable information, this individual of sample is such as human body, such as, three people 11,12 and 13 shown in Fig. 1.In actual enforcement, whom selects as individual of sample, can be foundation according to the motion of robot to be optimized.Such as, suppose to use robot to cook, that can select some cooks as individual of sample, or, also can be ordinary people not necessarily cook, but the motion of ordinary people when cooking will be selected to carry out gathering (follow-up will describe in detail); Again such as, suppose to use robot to carry out the workpiece handling of factory, some porters can be selected as individual of sample.
In addition, the quantity of selected individual of sample also can set flexibly, only exemplaryly in Fig. 1 shows three people, and the quantity in fact for being elected to be the people of individual of sample can be a lot, such as 50 people, 100 people.Under normal circumstances, the target data that sample size obtains more more meets the motion conditions of human body reality, such as, choose 10 porters and test the carrying action parameter obtained, with choose compared with carrying action parameter that 2 carrying work obtain, more easily will sum up the good action parameter of carrying action accurately.
As shown in Figure 1, be also provided with server 14, the data of above-mentioned individual of sample collection need to be sent to server 14, carry out process obtain target data by server 14 pairs of data.Such as, in above-mentioned example, the carrying action data of a lot of porters gathered is sent to server 14, server 14 according to these data analysis process, obtain these workmans general move action parameter, such as everybody carry time be all a angle that arm is raised up substantially, and lift b height, when even also may collect carrying, usual applied arm strength is c newton, is equivalent to some general character summing up most people motion, as the target data that will obtain.This target data transfers to robot 15 by server 14, and when robot 15 moves according to this target data, just can obtain good movement effects, robot is improved fast.
By foregoing description, as shown in Figure 2, server 14 can perform following flow process:
201, multiple movable informations of the corresponding multiple individual of sample of the difference of receiving sensor collection;
202, according to multiple movable information, the target travel information for optimizing the motion of movable information correspondence is obtained;
203, described target travel information is sent to artificial intelligence equipment, with the motion making described artificial intelligence equipment corresponding according to described target travel information and executing.
As shown in Figure 3, robot 15 can perform following flow process:
301, obtain target travel information, described target travel information is that motion control device obtains according to multiple movable informations of corresponding respectively multiple individual of sample of sensor collection;
302, corresponding according to described target travel information and executing motion, with by motion corresponding to multiple movable information described in described target travel Advance data quality.
Concrete, as follows by conjunction with the application scenarios of some reality or application mode, the process of above-mentioned server and robot is described in detail:
Application one: suppose robot be exclusively used in wash clothes (perhaps this example is not very appropriate, but only for illustration of the principle of the motion control method of the application), so, can choose and usually bear the more female individual of laundry work as individual of sample in family, and only gather the individual data when washing clothes.
Concrete, composition graphs 1, female individual arranges sensor, sensor installation 16 on individual 11, sensor installation 17 on individual 12, sensor installation 18 on individual 13.Due to mainly arm force of washing clothes, so can by these sensor settings on arm.The parameter of laundry action can comprise: dynamics, the amplitude of fluctuation of arm when giving the clothes a scrubbing etc. that arm applies when doing washing, can arrange the corresponding various sensors gathering these parameters.In addition, in order to not bring inconvenience to the life of test person, just can only install when washing clothes and start these sensors.
In the present embodiment, the data of sensor collection can be called movable information, namely to these some information of moving relevant of washing clothes; Can be passed to server 14 after this motion information acquisition, transfer mode does not limit, and can pass through network traffic.In server 14 side, will receive a lot of movable informations, the individual of sample that these information is respectively corresponding different, than three people as shown in Figure 1, everyone can transmit one to should the movable information of washing clothes that collects of test person.As following table 1 illustrates optional information recording method formula:
The movable information that table 1 sensor gathers
| Sample identification | Acquisition time | Dynamics information | Velocity information | Amplitude information | Environmental information |
| Y1 | 8:29 | N1 | V1 | B1 | S1 |
| Y2 | 16:40 | N2 | V2 | B2 | S2 |
| Y3 | 20:12 | N3 | V3 | B3 | S3 |
It should be noted that, be some optional information of example in table 1, can change in actual enforcement according to actual conditions.Ginseng is shown in Table 1, and Y1, Y2 and Y3 correspond respectively to the testing human of three shown in Fig. 1 11,12 and 13.Acquisition time can be test person start sensor when washing clothes, the start-up time of sensor, represents to start to wash clothes in this time; Optionally, this acquisition time also can be a time period, and 8:29 ~ 9:01 that such as Y1 is corresponding, 9:01 are wherein the time terminated of washing clothes.Velocity information wherein represents the swing speed of test person arm when washing clothes, and environmental information can be the humidity of the surrounding air that sensor gathers, and under normal circumstances, the environmental air humidity of washing clothes can be relatively high.In the present embodiment, such as, Y1 can be called an individual of sample, to the parameter combinations of (Y1,8:29, N1, V1, B1 and S1) of Y1 should being called the movable information of corresponding Y1 in so above-mentioned table 1.
Server 14 is after the above-mentioned multiple movable information getting sensor collection, the target travel information for optimizing the motion of movable information correspondence will be obtained according to these movable informations, such as, in above-mentioned example, acquire the kinematic parameter of washing clothes of a lot of test sample book, so what in the end wash clothes could clothes wash cleaner, such as arm needs the amplitude that swings much, the speed that arm swings is how many etc., and it is just more suitable that the Data Summary gathered by research goes out how to move.In concrete process, server can according to multiple body motion information (because movable information is the human body gathered, therefore the present embodiment is called body motion information) common feature, obtain the target body movable information of corresponding common feature, this target body movable information is exactly that can allowing of will finding is washed clothes the more excellent parameter of action.
Still illustrate for table 1: suppose that individual of sample have selected 10 people, in fact 10 parts of movable informations have been got in table 1, wherein, everyone has oneself kinematic parameter (such as speed, amplitude) when washing clothes, but server finds after statistics, and arm amplitude of fluctuation when having 7 people to wash clothes in these 10 people is at B
mleft and right, swing speed is at v
mleft and right, dynamics is at N
m, so these parameters (B
m, v
m, N
m) just can be called it is common feature, namely most people all so does, should (B
m, v
m, N
m) just can as target travel information.
As above, describe a kind of selection mode selecting 7 personal data from the sample data of 10 people, the mode of this searching common feature also can be understood as, and according to the data screening algorithm preset, removes insincere data from movable information; According to the trust data outside the insincere data in movable information, analyze and obtain described target travel information.Such as, in the data that these 10 people are corresponding, the swing speed of washing clothes is v1 respectively, v2, v3, v4 ... v10, wherein, the speed values (supposing it is identical speed unit) of v2 to v8 is supposed all about 15, have plenty of 14, have plenty of 16, have plenty of 15, and the numerical value of v1 is 6, the numerical value of v9 is 25, the numerical value of v10 is 10, so through statistics, object finds to meet the characteristics of motion of most people, then can by v1, the numerical value of v9 and v10 is removed, think that these are insincere data, be equivalent to abnormal data, only retain the speed values of v2 to v8, think trust data.And obtain target velocity numerical value according to the speed values of v2 to v8, such as can be set as 15, or be set as the range section of 14-16.And which kind of algorithm the present embodiment the data screening algorithm preset specifically adopts do not limit, as long as can realize the data screening function of above-mentioned principle.
Optionally, in actual test, everyone data may be not quite similar, duplicate multiple data can not be found, now server can according to certain algorithm, find most data to be in which type of scope, and then derive a common denominator data scope or certain point value within the scope of this is also passable.The target travel information that will obtain in a word is the parameter that most test sample book uses, can be seen by above-mentioned in addition, although in aforesaid description by the parameter combinations of (Y1,8:29, N1, V1, B1 and S1) as a movable information, the target travel information that obtain can be partial parameters (B wherein
m, v
m, N
m), can robot motion be used to guide.
This information, after obtaining target travel information, is sent to artificial intelligence equipment such as robot by server 14, the action that robot just can wash clothes according to this information and executing, such as according to (B
m, v
m, N
m) carry out swing and the force of robot arm.Because server passes to the target travel information of robot, obtain according to the actual motion Information Statistics of testing human, be parameter corresponding to action that most people all performs, the action that therefore robot performs according to this parameter can obtain reasonable effect; And the movable information of the present embodiment collection is not the data of indivedual individual of sample, but the large data of movable information corresponding to a lot of individual of samples, can add up and obtain good kinematic parameter.
In addition, the robot motion optimization of this mode, owing to being the common denominator data directly obtaining most sample, the speed that robot motion can be made to optimize improves, and compared to traditional isolated debugging, makes the acquisition efficiency comparison of more excellent data fast, and, automatically gather acquisition sample data by server in which, and robot is transferred to after statistics obtains target travel information, compared to traditional continuous amendment to robot program and debugging, the mode that robot is optimized is convenient and quick, such as from the angle of the user of robot, it no longer needs oneself debugging and amendment robot parameter, but obtain target travel information by robot oneself from server, this information is that network in charge acquisition and processing obtains, user is not needed to participate in, in a word, the motion control method of the present embodiment not only increases the optimal speed of robot, and owing to comparing the actual motion situation meeting individual of sample, effect of optimization will be better.
From the angle of robot 15, the target travel information that server process obtains can be stored in the memory module of robot 15; When not temporarily being connected by network between robot 15 with server 14, the up-to-date target travel information that the last time stored in memory module can obtain by robot is as the foundation of action executing, or, robot also can be connected by network with server, after server obtains up-to-date target travel information, perform corresponding action again.
Be exemplified below: user may not wish that the robot one of oneself is straight through network attached server, so user can at the 8:00 in morning of No. 1, start the machine the data acquisition functions of people, robot is networked and downloads up-to-date target travel information, according to the action that this information and executing is washed clothes from server.If user's sensation is good according to the effectiveness comparison of the information machine people motion that this is downloaded, this information can be used always; Or when user feels that robot motion is bad, clothes does not wash clean, when wanting to improve its movement effects further, the data acquisition functions of the people that can again start the machine, makes robot network from the target travel information of server down loading updating; Or, user also can control regularly from server obtaining information.Certainly optional, what server also can be regular pushes its target travel information upgraded according to Preset Time to robot, such as every other day pushes.
As above, the data acquisition and processing (DAP) of server side is not only carry out once, and the reception of data can be carried out always, or regularly carry out, constantly update to make target travel information.Concrete, such as, server can arrange a time interval, the calculating of performance objective movable information when the time interval of presetting arrives, and the 22:00 that such as can be set in every day, according to the movable information of the individual of sample gathered, calculates target travel information.Again such as, server also can the calculating of performance objective movable information when multiple movable information has upgraded, the individual of sample such as set is 10, so the collection movable information of these 10 individual of samples constantly updates, all can there be renewal possible every day, server, after reception one secondary data, when can wait until that the data of 10 all individual of samples have upgraded all again, performs above-mentioned process.Can certainly be other setting means, no longer illustrate.
Application two: in the example of above-mentioned application one, be that such as this intended application scene is washed clothes for certain specific intended application scene; Obtain the target travel information corresponding with scene of washing clothes, for optimizing the motion of washing clothes of robot.In the application two of the present embodiment, also can not distinguish scene, such as using someone as individual of sample for the moment, the various motions of this people in one day can be gathered, such as, wash clothes, cook, run, dance etc., gather the action of this test person always; Corresponding, robot receives target travel information also comprises information under a lot of scene, and the data of such as washing clothes, the data etc. of cooking, so this robot can be just the robot that can perform a lot of motor task.
Specifically describe as follows: for the test person 11 in Fig. 1, this test person can be arranged the various sensors for gathering its movable information, the set-up mode of sensor can set flexibly, makes test person carry comparatively for convenience and can not bring inconvenience to the action of test person as far as possible.Test person 11 in Fig. 1, only illustrates one of them sensor 16 that it is arranged with it, for gathering its arm movement, and can also at the sensor of other body parts setting for gathering corresponding parameter of this test person.Sensor can be always in running order, and such as can get up the time period come into play between sleep in evening from test person by day, sensor continually can carry out motion pick.The movable information gathered is similar to shown in table 1, such as can also comprise, the walking step pitch of test person, the speed of travel, environmental information can also comprise ambient air temperature etc., which kind of sensor is specifically set and gathers which kind of kinematic parameter, can determine according to the motion that robot is to be optimized.
In the present embodiment, first these data, after the above-mentioned data receiving sensor collection, can be distinguished according to application scenarios by server, and analyze corresponding target travel information to often kind of application scenarios respectively.Be exemplified below: with table 2 for exemplifying a part of acquisition parameter wherein.
The movable information that table 2 sensor gathers
As shown in table 2, suppose the movable information having 6 people, comprise multi-motion parameter; It may be corresponding different moving scenes in these movable informations, such as, when the sensor collecting test people movable information of a day, parameter when this test person is washed clothes is included, also comprising some actions during culinary art of this test person, also comprising this test person action when arranging housework etc.Different moving scenes, test person can take different mode of motion usually, and such as, when washing clothes, arm has amplitude of fluctuation and swing speed, and when hanging out, arm needs to raise, and the moving up and down when motion of arm may be frying pan cooking when cooking; These parameters can be sorted out according to moving scene by server, and data corresponding to same class scene just more easily find its common feature, and the data of inhomogeneity scene are normally not easy to find common feature.
For table 2, illustrate two kinds of moving scenes, wash clothes and cook, suppose that the movable information of 6 above-mentioned people is the information of corresponding these two kinds motions respectively, server can be distinguished according to environmental parameter.Such as, when washing clothes, air humidity can be relatively slightly higher, and ambient air temperature when cooking can be relatively high, based on this, humidity parameter (S1, S2 and S5) in the movable information of discovering server Y1, Y2 and Y5 is all in washes clothes in corresponding humidity range, all relatively high, accordingly the movable information of Y1, Y2 and Y5 is classified as " information that motion of washing clothes is corresponding ".Again such as, during culinary art, the air themperature in kitchen can be relatively high, based on this, temperature parameter (T3, T4 and T6) in the movable information of discovering server Y3, Y4 and Y6 is all in temperature range corresponding to air themperature, all relatively high, accordingly the movable information of Y3, Y4 and Y6 is classified as " information that culinary art motion is corresponding ".
As above, humidity parameter (S1, S2 and S5), temperature parameter (T3, T4 and T6) can as the foundations distinguishing moving scene, this kind of parameter can be called type of sports parameter, for determining type of sports, being such as wash clothes according to humidity determination type of sports, is culinary art according to temperature determination type of sports.According to these type of sports parameters, can from the multiple movable informations collected, acquisition belongs to the movable information of same type of sports as reference movement information, such as, Y1, Y2 and Y5 belong to motion of washing clothes, the movable information (such as dynamics, speed and amplitude etc.) that these samples are corresponding is called reference movement information, determines optimizing the target travel information of washing clothes and moving according to these reference movement information.After determining the reference movement information of same type of sports, according to the mode described, according to the data screening algorithm preset, insincere data can be removed from reference movement information above; According to the trust data outside the insincere data in reference movement information, analyze and obtain described target travel information.
Certainly above-mentionedly a kind of possible scene differentiating method is just illustrated, in concrete enforcement, server also can adopt additive method to carry out scene differentiation, such as, one can also be arranged for taking the equipment of test person sport video, the sport video of shooting is sent to server by this equipment, server can distinguish scene in conjunction with video analysis, such as server is by picture identification, obtain test person to cook, time when obtaining this culinary art is again 7:07, and then search the kinematic parameter that in movable information that this test person uploads, this time point is corresponding, and record scene corresponding to these parameters be culinary art.
Again such as, also can be controlled by test person, before certain moving scene of test person do-it-yourself (such as washing clothes), first from the sensory-control system that it carries, select corresponding scene (multiple scene option can be pre-set select for test person), to make, when sensor image data reports server, this scene information also to be transferred to server in the lump.Other scene is distinguished mode and is no longer itemized.
Server is after differentiation moving scene, and by respectively to the movable information of often kind of scene, find the target travel information that its common feature is corresponding, method detailed see the embodiment of application one, no longer can describe in detail.The target travel information of each scene corresponding is respectively sent to robot by server, and robot, according to these information, carries out correspondence according to parameter execution when performing different scene motions.
Such as, robot interior stores the motion foundation parameter of washing clothes and cooking, and when motion is washed clothes in robot execution, then therefrom obtains the parameter correspondence of washing clothes and performs an action, when robot performs culinary art motion, then the parameter correspondence therefrom obtaining culinary art performs an action.How robot distinguishes oneself on earth in which kind of motion of execution, several scenes recognition method can be had equally, such as, can there be controller in robot, when needs robot do which kind of motion time, can manually carry out selection setting by this controller, then robot recalls corresponding movable information execution accordingly from memory module; Again such as, robot also can pass through sensor senses surrounding environment, when such as humidity is higher, thinks to wash clothes, then selects the movable information of washing clothes, and thinks culinary art, then select the movable information of culinary art when temperature is higher.Equally, also can judge in conjunction with video.
Optionally, if server does not distinguish scene, also can take other data analysing method, find target travel information, to make the motion of robot, there is better effect.
Application three: in above-mentioned application one and application two, be all obtain the movable information of human body for example, the individual of sample of selection is people; In the application three of the present embodiment, the movable information of the individual of sample that server obtains is the movable information of robot, that is, individual of sample, except comprising test person, also comprises robot, and, also can carry out the optimization of common feature from the large data of the motion of robot collection, and re-enter robot after optimizing and carry out guidance machine people motion, form the close-loop feedback of robot motion optimization, so also make to promote further the optimal speed of robot motion, effect of optimization improves further.
Concrete, see Fig. 4, show the square ratio juris of the present embodiment, server 14 is after obtaining target body movable information according to the parameter gathered, this movable information can be sent to multiple robot, such as, comprise robot 15, robot 19 and robot 20, be only illustrate three robots in certain Fig. 4, can be a lot of robots in actual enforcement, to obtain the sample of large data.
That is, in the present embodiment, not only comprise people for the individual of sample tested, also comprise artificial intelligence equipment such as robot.Robot is also sensor is set, gather the movable information of this robot, such as, to wash clothes, the correlation parameter of washing clothes is transferred to robot 15 by server 14, robot 15 performs the action of washing clothes accordingly, in course of action, the sensor be arranged in robot starts to gather corresponding parameter, obtain robot some movable informations when washing clothes, the swing speed, amplitude of fluctuation etc. of the mechanical arm of such as robot, the collection of the movable information of robot is substantially identical with the method for the information acquisition of people, no longer describes in detail.If people is called body motion information as the parameter gathered during individual of sample, so robot can be called equipment moving information as the parameter gathered during individual of sample.
It should be noted that, the time that different robots obtains target travel information from server may not be synchronous, such as, the time 8:00 that robot 15 is set in every day obtains the target travel information upgraded from server 14, and the time 19:00 that robot 19 is set in every day obtains the target travel information upgraded from server 14, robot 20 does not have the set time, may be to obtain by the owner of robot (or being called user) control when hope obtains lastest imformation the target travel information upgraded from server 14.So this mode due to the information of server side be constantly update, therefore may be just different in the parameter of the action institute foundation of each robot of different time obtaining information, the movable information of the motion correspondence according to different information and executing can be gathered by sensor, feed back to server.
Server 14, after the equipment moving information receiving a lot of robot feedbacks, also can be taked the method identical with body motion information, analyze the common feature of these equipment moving information, and obtain corresponding target device movable information.Further, target body movable information and target device movable information can be carried out data coupling by server 14, the target travel information of the renewal comprehensively obtained both obtaining.
See Fig. 5, show the principle that server carries out data coupling, the specifically respective accounting respectively in movable information according to body motion information and equipment moving information, and target body movable information and target device movable information carry out data coupling, obtain target travel information, see following formula:
Target travel information=target body movable information * body motion information accounting+target device movable information * equipment moving information accounting;
Be exemplified below: for the arm swing speed of washing clothes during motion, suppose there are 20 personal accomplishment individual of samples, there is each self-corresponding swing speed respectively, and server analysis goes out target body movable information corresponding to these speed (i.e. target swing speed) is v1; In addition, suppose there are 10 robots as individual of sample, there is each self-corresponding swing speed respectively, and server analysis goes out target device movable information corresponding to these speed (i.e. target swing speed) is v2, so,
Body motion information accounting=20/ (20+10)=2/3;
Equipment moving information accounting=10/ (20+10)=1/3;
Target swing speed=v1* (2/3)+v2* (1/3)
Parameter in other movable informations can carry out the calculating of data coupling according to the method described above.In the movement control mode of the present embodiment, the large data gathered from a fairly large number of individual of sample that server obtains, both the data gathered from human body had been comprised, also the data gathered from robot are comprised, and carry out being coupled according to these two parts data and obtain new target travel information, server can by this new target travel information transmission to robot, to optimize the action of robot further.
Application four: in above-mentioned application three, server combines body motion information and equipment moving information, comprehensively obtains the target travel information being used to guide robot motion; In the application four of the present embodiment, improve further on this basis, on the basis that server is optimized according to above-mentioned two kinds of information, progressively reduce the proportion shared by body motion information, and final self-teaching and the closed loop training forming robot self.
Optionally, server according to the information regularization condition preset, can adjust body motion information, to make the accounting of body motion information in total movable information reduce, namely progressively reduces the proportion shared by body motion information.The above-mentioned information regularization condition preset, be such as, server just reduces the proportion of body motion information every a week, such as the proportion of the first week body motion information in total information is 70%, the proportion of second week adjustment body motion information in total movable information is 50%, etc.; Or this condition can also arrange the proportion how reducing body motion information, such as reduces by 10% at set intervals, or reduction by 5% etc., various ways can set flexibly.
Wherein, the accounting of body motion information reduces, following mode can be adopted: the quantity that can gradually reduce the testing human as individual of sample, such as, have selected 20 personal accomplishment individual of samples at first, through after a period of time, eliminate 5 people, change the individual of sample of 15 people into, then through being reduced to the individual of sample comprising 10 people after a while, gradually reduce until the final individual of sample cancelling human body.Or the quantity that also can be the individual of sample gathered is constant, just server is when analyzing and processing data, and the data of the individual of sample of use gradually reduce.When body motion information is reduced to zero, the movable information that the individual of sample that server receives gathers, only has the equipment moving information gathered from robot by sensor, server is according to these equipment moving information searching common features, obtain target device movable information, and feed back to robot, robot performs corresponding motion again according to this target device movable information, feed back to server again after collecting movable information, define the closed loop self-teaching of robot data.
In addition, in the process of the self-closed loop training of robot, if the user of robot thinks that the action of robot can not meet the requirement of user, user also can modify to the kinematic parameter of robot a little, the movable information of the action that such correspondence performs also can change, after transferring to server, find common feature by server according to these movable informations, the action of further guidance machine people.
In the motion control method of the embodiment of the present application, not only the amendment of single robot motion's parameter is debugged, but the movable information sample of the large data gathered by the automatic collecting sensor of server, and obtain preferably kinematic parameter according to these large data message samples searching common features, the action of guidance machine people is carried out with this, the optimization efficiency of robot is improved, and effect of optimization is also better.
In addition, it should be noted that, motion performed by robot, can be not only wash clothes, cook, or performing certain factory's task dispatching can help the mankind to process the action of certain task, also can be the robot for amusement, the robot of such as dancing, or imitate the mankind and carry out robot of running etc., these robots mankind are also wished that its motion done can closer to the mankind, because the mankind can using the partner of these robots as a daily life of the mankind like this, accompany mankind's amusement or motion etc. together, also be helpful to the life of the mankind in fact.In this case, just can adopt the method described in above-described embodiment, select human body as individual of sample, gather the movable information that a lot of human body is danced, server counts the action parameter adopted when most of human body is danced accordingly, guidance machine people moves, and robot is danced more as the mankind.Further, in the motion control method of the embodiment of the present application, the large data of collection of server constantly gather to constantly update, and makes the parameter being used to guide motion also constantly be optimized like this.
On the basis of above-mentioned account for motion control method, the embodiment of the present application additionally provides a kind of motion control device, and this device can be arranged on the server 14, performs motion control method corresponding in said method embodiment to make server.See Fig. 6, this motion control device can comprise: information receiving module 61, message processing module 62 and information sending module 62; Wherein,
Information receiving module 61, for multiple movable informations that receiving sensor gathers, described multiple movable information corresponding multiple individual of sample respectively;
Message processing module 62, for according to described multiple movable information, determines the target travel information optimizing corresponding motion;
Information sending module 63, for described target travel information is sent to artificial intelligence equipment, with the motion making described artificial intelligence equipment corresponding according to described target travel information and executing.
Further, message processing module 62, specifically for according to the type of sports parameter in described movable information, determines the type of sports that described movable information is corresponding; From described multiple movable information, acquisition belongs to the movable information of same type of sports as reference movement information, and determines according to described reference movement information the target travel information optimizing corresponding motion.
Further, described individual of sample comprises: human body; Described movable information comprises: the body motion information that the described human body that sensor gathers is corresponding.Described individual of sample also comprises: smart machine; Described movable information also comprises: the equipment moving information that the smart machine that sensor gathers is corresponding.Message processing module 62, specifically for according to body motion information determination target body movable information; According to equipment moving information determination target device movable information; And according to described target body movable information, target device movable information and described body motion information and the respective accounting of equipment moving information respectively in described movable information, carry out data coupling, obtain described target travel information.
The embodiment of the present application additionally provides a kind of artificial intelligence equipment, and this equipment is such as robot; See Fig. 7, this artificial intelligence equipment can comprise: data obtaining module 71 and Motor execution module 72; Wherein,
Data obtaining module 71, for obtaining target travel information, described target travel information is that motion control device obtains according to multiple movable informations of sensor collection, described multiple movable information corresponding multiple individual of sample respectively;
Motor execution module 72, for the motion corresponding according to described target travel information and executing, with by motion corresponding to multiple movable information described in described target travel Advance data quality.
Optionally, data obtaining module 71, specifically for: obtain and be stored in advance in local described target travel information; Or, obtain described target travel information from described motion control device.
Please refer to Fig. 8, the embodiment of the present application additionally provides a kind of entity structure schematic diagram of motion control device.This motion control device may be the host server comprising computing power, or personal computer, or portable portable computer or terminal etc., and the present embodiment does not limit the specific implementation of this device.As shown in Figure 8, this motion control device can comprise: processor (processor) 810, communication interface (CommunicationsInterface) 820, storer (memory) 830 and bus 840.
Wherein, processor 810, communication interface 820, storer 830 completes mutual communication by bus 840.Communication interface 820, carries out the transmission of movable information for carrying out communicating with the communication interface in sensor or robot.
Processor 810 may be a central processor CPU, or specific integrated circuit ASIC (ApplicationSpecificIntegratedCircuit), or is configured to the one or more integrated circuit implementing the present embodiment.Storer 830, may comprise high-speed RAM storer, still may comprise nonvolatile memory (non-volatilememory), such as at least one magnetic disk memory, and this storer 830 is for depositing programmed instruction.Processor 810 can programmed instruction in execute store 830, for performing following work: multiple movable informations of the corresponding multiple individual of sample of the difference that receiving sensor gathers; According to described multiple movable information, obtain the target travel information for optimizing the motion of described movable information correspondence; Described target travel information is sent to artificial intelligence equipment, with the motion making described artificial intelligence equipment corresponding according to described target travel information and executing.
Please refer to Fig. 9, the embodiment of the present application additionally provides a kind of entity structure schematic diagram of artificial intelligence equipment.This manual's smart machine is such as robot.As shown in Figure 9, this artificial intelligence equipment can comprise: processor (processor) 910, communication interface (CommunicationsInterface) 920, storer (memory) 930 and bus 940.
Wherein, processor 910, communication interface 920, storer 930 completes mutual communication by bus 940.Communication interface 920, carries out the transmission of movable information for carrying out communicating with server.
Processor 910 may be a central processor CPU, or specific integrated circuit ASIC (ApplicationSpecificIntegratedCircuit), or is configured to the one or more integrated circuit implementing the present embodiment.Storer 930, may comprise high-speed RAM storer, still may comprise nonvolatile memory (non-volatilememory), such as at least one magnetic disk memory, and this storer 930 is for depositing programmed instruction.Processor 910 can programmed instruction in execute store 930, for the following work of execution: obtain target travel information, described target travel information is that motion control device obtains according to multiple movable informations of corresponding respectively multiple individual of sample of sensor collection; The motion corresponding according to described target travel information and executing, with by motion corresponding to multiple movable information described in described target travel Advance data quality.
Those skilled in the art can be well understood to, and for convenience and simplicity of description, the specific works process of the system of foregoing description, device and unit, with reference to the corresponding process in preceding method embodiment, can not repeat them here.
In several embodiments that the application provides, should be understood that disclosed system, apparatus and method can realize by another way.Such as, device embodiment described above is only schematic, such as, the division of described unit, be only a kind of logic function to divide, actual can have other dividing mode when realizing, such as multiple unit or assembly can in conjunction with or another system can be integrated into, or some features can be ignored, or do not perform.Another point, shown or discussed coupling each other or direct-coupling or communication connection can be by some communication interfaces, and the indirect coupling of device or unit or communication connection can be electrical, machinery or other form.
The described unit illustrated as separating component or can may not be and physically separates, and the parts as unit display can be or may not be physical location, namely can be positioned at a place, or also can be distributed in multiple network element.Some or all of unit wherein can be selected according to the actual needs to realize the object of the present embodiment scheme.
In addition, each functional unit in each embodiment of the present invention can be integrated in a processing unit, also can be that the independent physics of unit exists, also can two or more unit in a unit integrated.
If described function using the form of SFU software functional unit realize and as independently production marketing or use time, can be stored in a computer read/write memory medium.Based on such understanding, the part of the part that technical scheme of the present invention contributes to prior art in essence in other words or this technical scheme can embody with the form of software product, this computer software product is stored in a storage medium, comprising some instructions in order to make a computer equipment (can be personal computer, server, or the network equipment etc.) perform all or part of step of method described in each embodiment of the present invention.And aforesaid storage medium comprises: USB flash disk, portable hard drive, ROM (read-only memory) (ROM, Read-OnlyMemory), random access memory (RAM, RandomAccessMemory), magnetic disc or CD etc. various can be program code stored medium.
The foregoing is only preferred embodiment of the present invention, not in order to limit the present invention, within the spirit and principles in the present invention all, any amendment made, equivalent replacement, improvement etc., all should be included within the scope of protection of the invention.
Claims (16)
1. a motion control method, is characterized in that, comprising:
Obtain multiple movable informations that sensor gathers, described multiple movable information corresponding multiple individual of sample respectively;
According to described multiple movable information, determine the target travel information optimizing corresponding motion;
Described target travel information is sent to artificial intelligence equipment, with the motion making described artificial intelligence equipment corresponding according to described target travel information and executing.
2. method according to claim 1, is characterized in that, described according to described multiple movable information, determines the target travel information optimizing corresponding motion, is specially:
When the time interval of presetting arrives, or when described multiple movable information has upgraded, according to described multiple movable information, calculate the target travel information for optimizing corresponding motion.
3. method according to claim 1, is characterized in that, described according to described multiple movable information, determines the target travel information optimizing corresponding motion, comprising:
According to the type of sports parameter in described movable information, determine the type of sports that described movable information is corresponding;
From described multiple movable information, acquisition belongs to the movable information of same type of sports as reference movement information, and determines according to described reference movement information the target travel information optimizing corresponding motion.
4. method according to claim 3, is characterized in that, the described target travel information determining optimizing corresponding motion according to described reference movement information, comprising:
According to the data screening algorithm preset, from described reference movement information, remove insincere data;
According to the trust data outside the insincere data in described reference movement information, analyze and obtain described target travel information.
5., according to the arbitrary described method of Claims 1 to 4, it is characterized in that, described individual of sample comprises: human body; Described movable information comprises: the body motion information that the described human body that sensor gathers is corresponding.
6. method according to claim 5, is characterized in that, described individual of sample also comprises: smart machine; Described movable information also comprises: the equipment moving information that the described smart machine that sensor gathers is corresponding; Described according to described multiple movable information, obtaining the target travel information for optimizing corresponding motion, comprising:
According to body motion information determination target body movable information;
According to equipment moving information determination target device movable information;
According to described target body movable information, target device movable information and described body motion information and the respective accounting of equipment moving information respectively in described movable information, carry out data coupling, obtain described target travel information.
7. method according to claim 6, is characterized in that, described according to body motion information determination target body movable information before, also comprise:
According to the information regularization condition preset, described body motion information is adjusted, reduce to make the accounting of described body motion information in described movable information.
8. a motion control method, is characterized in that, comprising:
Obtain target travel information, described target travel information is that motion control device obtains according to multiple movable informations of sensor collection, described multiple movable information corresponding multiple individual of sample respectively;
The motion corresponding according to described target travel information and executing, with by motion corresponding to multiple movable information described in described target travel Advance data quality.
9. method according to claim 8, is characterized in that, described acquisition target travel information, comprising: obtain and be stored in advance in local described target travel information; Or, obtain described target travel information from described motion control device.
10. according to the method that claim 9 is stated, it is characterized in that, describedly obtain described target travel information from described motion control device, comprising:
Receive the described target travel information upgraded from described motion control device according to Preset Time.
11. 1 kinds of motion control devices, is characterized in that, comprising:
Information receiving module, for multiple movable informations that receiving sensor gathers, described multiple movable information corresponding multiple individual of sample respectively;
Message processing module, for according to described multiple movable information, determines the target travel information optimizing corresponding motion;
Information sending module, for described target travel information is sent to artificial intelligence equipment, with the motion making described artificial intelligence equipment corresponding according to described target travel information and executing.
12. devices according to claim 11, is characterized in that,
Described message processing module, specifically for according to the type of sports parameter in described movable information, determines the type of sports that described movable information is corresponding; From described multiple movable information, acquisition belongs to the movable information of same type of sports as reference movement information, and determines according to described reference movement information the target travel information optimizing corresponding motion.
13. devices according to claim 11, is characterized in that, described individual of sample comprises: human body; Described movable information comprises: the body motion information that the described human body that sensor gathers is corresponding.
14. devices according to claim 13, is characterized in that, described individual of sample also comprises: smart machine; Described movable information also comprises: the equipment moving information that the smart machine that sensor gathers is corresponding;
Described message processing module, specifically for according to body motion information determination target body movable information; According to equipment moving information determination target device movable information; And according to described target body movable information, target device movable information and described body motion information and the respective accounting of equipment moving information respectively in described movable information, carry out data coupling, obtain described target travel information.
15. 1 kinds of artificial intelligence equipment, is characterized in that, comprising:
Data obtaining module, for obtaining target travel information, described target travel information is that motion control device obtains according to multiple movable informations of sensor collection, described multiple movable information corresponding multiple individual of sample respectively;
Motor execution module, for the motion corresponding according to described target travel information and executing, with by motion corresponding to multiple movable information described in described target travel Advance data quality.
16. artificial intelligence equipment according to claim 15, is characterized in that,
Described data obtaining module, specifically for: obtain and be stored in advance in local described target travel information; Or, obtain described target travel information from described motion control device.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810672928.2A CN109048889B (en) | 2014-09-10 | 2014-09-10 | Method and device for obtaining target motion information of artificial intelligence equipment |
| CN201410459517.7A CN105467926B (en) | 2014-09-10 | 2014-09-10 | A kind of motion control method and device, artificial intelligence equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410459517.7A CN105467926B (en) | 2014-09-10 | 2014-09-10 | A kind of motion control method and device, artificial intelligence equipment |
Related Child Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810672928.2A Division CN109048889B (en) | 2014-09-10 | 2014-09-10 | Method and device for obtaining target motion information of artificial intelligence equipment |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105467926A true CN105467926A (en) | 2016-04-06 |
| CN105467926B CN105467926B (en) | 2018-08-07 |
Family
ID=55605737
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810672928.2A Active CN109048889B (en) | 2014-09-10 | 2014-09-10 | Method and device for obtaining target motion information of artificial intelligence equipment |
| CN201410459517.7A Active CN105467926B (en) | 2014-09-10 | 2014-09-10 | A kind of motion control method and device, artificial intelligence equipment |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810672928.2A Active CN109048889B (en) | 2014-09-10 | 2014-09-10 | Method and device for obtaining target motion information of artificial intelligence equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN109048889B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111383059A (en) * | 2017-12-28 | 2020-07-07 | 纳恩博（北京）科技有限公司 | Information processing method and device, electronic device and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1830635A (en) * | 2005-03-10 | 2006-09-13 | 北京理工大学 | Human imitation robot action similarity evaluation based on human body motion track |
| DE102006017945A1 (en) * | 2005-04-15 | 2006-12-14 | Fanuc Robotics America, Inc., Rochester Hills | Robot control system, has personal computer connected with robot control and robot power data monitoring unit so that computer responds to power data to use optimization routines and to change robot control system parameter in real time |
| JP2009502527A (en) * | 2005-08-01 | 2009-01-29 | トヨタ自動車株式会社 | Robot control system and robot control method |
| CN103605375A (en) * | 2013-10-24 | 2014-02-26 | 国家电网公司 | Bionic robot control method and bionic robot control device |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101266619B (en) * | 2008-05-12 | 2010-07-21 | 腾讯科技（深圳）有限公司 | User information excavation method and system |
-
2014
- 2014-09-10 CN CN201810672928.2A patent/CN109048889B/en active Active
- 2014-09-10 CN CN201410459517.7A patent/CN105467926B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1830635A (en) * | 2005-03-10 | 2006-09-13 | 北京理工大学 | Human imitation robot action similarity evaluation based on human body motion track |
| DE102006017945A1 (en) * | 2005-04-15 | 2006-12-14 | Fanuc Robotics America, Inc., Rochester Hills | Robot control system, has personal computer connected with robot control and robot power data monitoring unit so that computer responds to power data to use optimization routines and to change robot control system parameter in real time |
| JP2009502527A (en) * | 2005-08-01 | 2009-01-29 | トヨタ自動車株式会社 | Robot control system and robot control method |
| CN103605375A (en) * | 2013-10-24 | 2014-02-26 | 国家电网公司 | Bionic robot control method and bionic robot control device |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111383059A (en) * | 2017-12-28 | 2020-07-07 | 纳恩博（北京）科技有限公司 | Information processing method and device, electronic device and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109048889B (en) | 2021-03-23 |
| CN105467926B (en) | 2018-08-07 |
| CN109048889A (en) | 2018-12-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102184454B (en) | Granulator formula generation method based on neural network system | |
| CN110689535B (en) | Workpiece identification method and device, electronic equipment and storage medium | |
| CN105184367B (en) | The model parameter training method and system of deep neural network | |
| Manju et al. | An analysis of Q-learning algorithms with strategies of reward function | |
| CN104007822A (en) | Large database based motion recognition method and device | |
| CN109358546B (en) | Control method, device and system of household appliance | |
| CN108983979A (en) | A kind of gesture tracking recognition methods, device and smart machine | |
| CN104090189A (en) | Device working state detecting method and device | |
| CN1224515A (en) | Adaptive object-oriented optimization software system | |
| CN109460792A (en) | A kind of artificial intelligence model training method and device based on image recognition | |
| CN109063790A (en) | Object identifying model optimization method, apparatus and electronic equipment | |
| CN105467926A (en) | Motion control method and device and artificial intelligence device | |
| CN110046585A (en) | A kind of gesture identification method based on environment light | |
| CN105528615B (en) | The optimum path search method of behavioral data | |
| Li et al. | Predicting thread profiles across core types via machine learning on heterogeneous multiprocessors | |
| CN111339838A (en) | Pig behavior identification method and system based on information fusion | |
| CN109977884A (en) | Target follower method and device | |
| Leye et al. | GUISE-a tool for GUIding simulation experiments | |
| CN113988562A (en) | Intelligent household energy consumption management method and device, storage medium and management equipment | |
| CN113792435A (en) | Data model establishing method and device and server | |
| CN112613184A (en) | Artificial intelligence method for judging distance of side slope collapse rockfall during earthquake occurrence | |
| CN111428224A (en) | Computer account login method based on face recognition | |
| CN109325442A (en) | A kind of face identification method of image pixel missing | |
| Ling et al. | An adaptive parameter control strategy for ant colony optimization | |
| Xie et al. | A LoRa-based remote gesture monitoring system using deep learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20200923 Address after: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman Islands Patentee after: Innovative advanced technology Co.,Ltd. Address before: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman Islands Patentee before: Advanced innovation technology Co.,Ltd. Effective date of registration: 20200923 Address after: Cayman Enterprise Centre, 27 Hospital Road, George Town, Grand Cayman Islands Patentee after: Advanced innovation technology Co.,Ltd. Address before: A four-storey 847 mailbox in Grand Cayman Capital Building, British Cayman Islands Patentee before: Alibaba Group Holding Ltd. |