CN1916969A - Method for generating reaction accompany movement based on hybrid control - Google Patents

Method for generating reaction accompany movement based on hybrid control Download PDF

Info

Publication number
CN1916969A
CN1916969A CNA2006100528111A CN200610052811A CN1916969A CN 1916969 A CN1916969 A CN 1916969A CN A2006100528111 A CNA2006100528111 A CN A2006100528111A CN 200610052811 A CN200610052811 A CN 200610052811A CN 1916969 A CN1916969 A CN 1916969A
Authority
CN
China
Prior art keywords
motion
sequence
reaction
physical simulation
visual human
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
Application number
CNA2006100528111A
Other languages
Chinese (zh)
Inventor
潘志庚
唐冰
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Zhejiang University ZJU
Original Assignee
Zhejiang University ZJU
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Zhejiang University ZJU filed Critical Zhejiang University ZJU
Priority to CNA2006100528111A priority Critical patent/CN1916969A/en
Publication of CN1916969A publication Critical patent/CN1916969A/en
Pending legal-status Critical Current

Links

Landscapes

  • Processing Or Creating Images (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

A method for generating follow-up motion of reaction based on mixed control includes predicting motion trace of virtual person under physical simulation according to received foreign event, searching out relevant motion sequence from motion-catching databank, selecting out transition sequence of next stage to generate follow-up motion of reaction based on physical means, connecting two motion sequences before and after said event and regulating them to make physically simulated motion have self-protection and self-balance consciousness of human being under agravity condition.

Description

A kind of based on the reaction accompany movement generation method of mixing control
Technical field
The invention belongs to the computer animation field, especially relate to a kind of based on the reaction accompany movement generation method of mixing control.
Technical background
Along with the widespread use of human body animation in virtual reality, 3d gaming, digital entertainment, people have higher requirement to the sense of reality and the interactivity of human body animation.The reaction followability human body animation that generation is made really, can be controlled unexpected incident is the challenging work in human body animation field always.Keep physical constraint, import or the like owing to reaction followability human body animation requires visual human's motion usually, therefore react the followability human body animation its generation technique is had higher requirement with environmental interaction nature, real-time response user.Reaction of formation followability human body animation mainly contains and adopts data-driven version or physical simulating method at present.
Data-driven version mainly refers to the motion capture data type of drive.Motion capture data drives the visual human by writing down real human body movement data, therefore comprises rich details information, advantages such as the animation strong sense of reality that have high-level efficiency, is generated.But with regard to its technology itself, the capturing movement technology is that the data of being obtained can only be used under particular environment to duplicating of moving.Because data-driven version too relies on the motion sequence of catching in advance or generating, they not too are fit to have the external force effect or with environment mutual, unpredictable scene are closely arranged.And need in the entire database scope, carry out similarity and calculate, cost is very high, generally can not make a response to user's input in real time.Therefore only specific reaction campaign is simulated usually based on the method for data-driven, for example only keep reaction to handle, can reduce the data and the search volume of catching like this at the balance of making after being " pushed ".
The most natural a kind of mode of personage and environment dynamic interaction of handling is exactly to adopt the method for physical simulation.Recently, in having the recreation of advanced physical simulation special efficacy, some adopt the Rag-doll system to come reciprocation between the virtual human and environment of dynamic similation.Generate the behavior with human characteristics owing to lack the necessary control system, they are normally passive to the response of external force, so the Rag-doll simulation seems to have no life.Though some researchers attempt to design some Dynamics Controllers guide the visual human realize living in some daily activities, as walking, balance, fall down and roll ground etc.Because the complicacy of Dynamics Controller design and lack understanding to the human motion physical rules is if only use dynamics methods, even look that very simple task (for example, along the straight line walking) all is very difficult to realize.
Aforesaid motion generating algorithm in use all has certain limitation, is difficult to satisfy generate the needs that the visual human reacts the followability action in real time.Such as, the calculated amount of these algorithms is all very big, can't allow the user obtain the reaction effect that the visual human disturbs to external world in real time.Some method can only be handled special input mode, such as pushing away and drawing, is difficult to accomplish the accuracy on the details.
Summary of the invention
The purpose of this invention is to provide a kind of based on the reaction accompany movement generation method of mixing control.
The step that the present invention generates method is as follows:
1) according to the external event of accepting, the movement locus of prediction visual human under physical simulation, the steps include: that (1) create two identical scenes simultaneously, each scene is under the jurisdiction of a physical simulation process, (2) time step of prognosis modelling process is made as ten times of common simulation, (3) prognosis modelling process and common simulation process are set to start to walk at one time, (4) motion sequence that prediction is come out and the motion sequence in the database compare, seek suitable redirect sequence and best redirect opportunity, and feed back to normal physical simulation process;
2) adopt approximate nearest neighbor search algorithm (ANN, Approximate NearestNeighbors, this method is the conventional algorithm of neighbor search) carry out motion search, and the method for the level of detail reduction search volume dimension of employing skeletal structure improves the speed of motion search;
3) making similarity between any two frames to motion sequence in movement locus under the physical simulation and the database calculates, difference between sequence that physical simulation obtains and two frames of the sequence in the database is expressed as the position in corresponding joint and the weights sum of angle difference.In order to guarantee the similarity of joint on speed, and guarantee between going To Frame, too big difference to be arranged,, can expand in the window size scope calculation of similarity degree.The size of window is chosen as a needed time of typical redirect.Suppose f sAnd f mRepresent the movement locus of prediction and two frames in the motion sequence in the database respectively, then in a window, two frame difference formula can be expressed as:
Figure A20061005281100071
Wherein ws is the size of redirect window, and J is visual human's a joint quantity.p j(f Si) expression f SiThe position in j joint of frame, similarly θ j(f Si) be f SiThe position angle in j joint of frame.T θ, l(f S0, f M0) be a coordinate transform operation.According to the health different parts similarity is calculated the size of contribution, w PjAnd w θ jRespectively to the linear distance and the angular distance weighting in j joint;
4) employing moves the linking incident that forward and backward these two sections motion sequences take place based on the method reaction of formation followability of physics, the steps include: that (1) carry out interpolation with original series and the sequence that searches, obtain joint attitude angle θ after two sections sequence interpolation before and after redirect Blend(t), (2) in the process of physical simulation, the interpolating sequence of Dynamics Controller two sections sequences along front and back generates corresponding reaction action; (3) for each joint, Dynamics Controller adopts proportion differential servomechanism calculating effect moment thereon:
τ = k s ( θ blend ( t ) - θ cur ) - k d ( - θ · cur )
K wherein sAnd k dBe respectively the rigidity and the damping gain of corresponding joint, θ CurWith Represent current joint attitude angle and angular velocity respectively.
5) adopt biomechanical principle that the reaction accompany movement that generates is adjusted; the steps include: that the moment that (1) adopts proportional plus derivative controller to produce will drive the top limbs continuously to the T attitude; (2) the controller drives both arms are followed the tracks of the touchdown point of health, the action that generates protectiveness is hindered avoiding, and (3) adopt the controller drives visual human to arrive or near the initial attitude of motion sequence that will redirect.
Said establishment physical simulation scene refers to: set up visual human's dynamics human skeleton model, be associated with the manikin of hinge arrangement catching the data of getting off, set up hybrid-type visual human's skeleton.In order to reduce complexity of dynamic model, 15 joints only from catch data, have been extracted.The size of each part of visual human's health is determined by the skeleton of catching in the data.The quality main reference of health each several part is from the measurement data in the anatomy pertinent literature, and made corresponding convergent-divergent according to the skeleton size of catching in the data.Visual human's motion is driven by corresponding motion capture data, and the physical attribute of each part of health can be inquired about at any time in motion process.
The step of said Parallel Simulation mechanism is: 1) when because collision or foreign impacts when changing visual human's state, physical simulation process that has big time step is used for apace predicated response result and obtain the track of visual human in this physical simulation process, 2) this is the result of visual human's passive response external event with the motion sequence that the physical simulation process generates fully, be mainly used to search for suitable redirect motion sequence, 3) simultaneously, the motion control that another physical simulation process with step-length normal time will be taken over the visual human, and generation is to the reaction action of external event.
Said motion selection method is: the motion sequence in initial analog track (movement locus that prediction is come out) and the database is compared, rather than do similarity and calculate as commonsense method between two sections motion sequences that existed.
Said approximate nearest neighbor search algorithm is: according to the error boundary ε and the data query q of user's appointment, and the data p that the space tree search that utilization builds meets the following conditions:
dist(p，q)≤(1+ε)dist(p *，q)
P wherein *Be the real neighbour of q.
The method that the level of detail of said skeletal structure reduces the search volume dimension is: set up the hierarchical model that human body closes skeleton, according to stratum level, reduce the articulation point number successively.According to search efficiency, by selecting suitable level skeletal structure, system compromises on speed and moving-mass automatically.
The present invention proposes and physically react the method that followability is moved really a kind of can in motion capture data, the adding: according to the external event of accepting, from the motion capture data storehouse, search out the motion sequence that adapts with it, employing is based on the motion of the method reaction of formation followability of physics, and is used for the linking incident forward and backward these two sections motion sequences take place.In order accurately to predict visual human's movement locus, the present invention proposes the method for Parallel Simulation.Method of the present invention can be simulated the motion falling down, get up after the visual human runs into barrier continuously and in real time by the actual response of making to hand push, after beating.In the physical simulation process, the invention allows for attitude adjustment based on biomechanics, the strategy that adopts arm to absorb instant impact is simulated the protectiveness action in the process of falling down.In the unbalanced process of visual human, because the adjustment of process biomechanics characteristic, the present invention generates the more approaching true people of action.
Method of the present invention adopts the movement locus prediction based on Parallel Simulation mechanism, can predict the movement locus of visual human under dynamics Controlling exactly, guarantees the flatness of motion transition.The present invention has adopted the method for continuous attitude control, makes Dynamics Controller drive the visual human along the interpolation exercise data that the two sections sequences in front and back take place in external disturbance, and generates corresponding continuity reaction action.The protectiveness action with human behavior characteristics is simulated in the attitude adjustment based on biomechanics that the present invention proposes, the strategy that adopts arm to absorb the impulsive force of moment, makes the motion of virtual human of generation more true to nature.Therefore the present invention can fast and effeciently simulate the visual human to continuously, the repeatedly reaction action of trigger event, and this method uses Parallel Simulation mechanism prediction visual human's movement locus, can accomplish in real time.This method can allow user and visual human to carry out alternately, has good realization and is worth.
Embodiment
Principle based on the reaction accompany movement generation method of mixing control is: these two kinds technology that generate human body animation of physical simulation and capturing movement have certain complementarity.The method that this method adopts data-driven to combine with dynamics Controlling improves the physical third dimension of human body animation.Disturb according to the external force that the visual human is subjected to, the movement locus of prediction visual human under physical simulation, from the motion capture data storehouse, search out the motion sequence that adapts with it then, and adopt method reaction of formation followability motion, and be used for the linking incident forward and backward these two sections motion sequences take place based on physics.That is to say, when the visual human with extraneous when mutual, adopt and catch data-driven; When the visual human is subjected to the external force effect, jump to the physical simulation state, resolve the generation corresponding action by dynamics.When the external force action effect disappears, when physical simulation does not have motion capture data reliable, select suitable frame sequence to get back to the motion capture data driving condition again.Wherein, the prediction of visual human's movement locus under physical simulation is the key of reaction of formation campaign, the present invention proposes the mechanism that adopts Parallel Simulation for this reason.Step is:
1) according to the external event of accepting, the movement locus of prediction visual human under physical simulation the steps include: that (1) create two identical scenes simultaneously, and each scene is under the jurisdiction of a physical simulation process.Wherein the foundation of the kinetic model of visual human's skeleton is the key of creating the physics scene.Be associated with the manikin of hinge arrangement catching the data of getting off, set up hybrid-type visual human's skeleton.In order to reduce complexity of dynamic model, 15 joints only from catch data, have been extracted.The size of each part of visual human's health is determined by the skeleton of catching in the data.The quality main reference of health each several part is from the measurement data in the anatomy pertinent literature, and made corresponding convergent-divergent according to the skeleton size of catching in the data.Each human synovial can be regarded as a rotation or spherical joint, has one or three degree of freedom respectively.
(2) time step of prognosis modelling process is made as ten times of common simulation.
(3) Parallel Simulation, prognosis modelling process and common simulation process are set to start to walk at one time.The final motion sequence that physical simulation generates with original motion sequence and next step will redirect motion sequence be connected.
(4) motion sequence that prediction is come out and the motion sequence in the database compare.
(5) seek suitable redirect sequence and best redirect opportunity, and feed back to normal physical simulation process.
2) rapid movement search, the level of detail of approximate nearest neighbor search algorithm (ANN, ApproximateNearest Neighbors) of employing and skeletal structure reduces the method for search volume dimension, improves the speed of motion search.Make up a kd tree or bbd tree (balanced box-decomposition tree), subdivision is carried out in the space.According to the error boundary ε and the data query q of user's appointment, the data p that the space tree search that utilization builds meets the following conditions:
dist(p，q)≤(1+ε)dist(p *，q)
P wherein *Be the real neighbour of q;
The method that the level of detail of said skeletal structure reduces the search volume dimension is: set up the hierarchical model that human body closes skeleton, according to stratum level, reduce the articulation point number successively.According to search efficiency, by selecting suitable level skeletal structure, system compromises on speed and moving-mass automatically.
3) motion is selected, and makes similarity between any two frames to motion sequence in movement locus under the physical simulation and the database and calculates.Like this, the difference between sequence that physical simulation obtains and two frames of the sequence in the database can be expressed as the position in corresponding joint and the weights sum of angle difference.In order to guarantee the similarity of joint on speed, and guarantee between going To Frame, too big difference to be arranged,, can expand in the window size scope calculation of similarity degree.The size of window is chosen as a needed time of typical redirect.Suppose f sAnd f mRepresent the movement locus of prediction and two frames in the motion sequence in the database respectively, then in a window, two frame difference formula can be expressed as:
Figure A20061005281100101
Wherein ws is the size of redirect window, and J is visual human's a joint quantity.p j(f Si) expression f SiThe position in j joint of frame, similarly θ j(f Si) be f SiThe position angle in j joint of frame.T θ, l(f S0, f M0) be a coordinate transform operation.According to the health different parts similarity is calculated the size of contribution, w PjAnd w θ jRespectively to the linear distance and the angular distance weighting in j joint;
4) employing is based on the method reaction of formation followability motion of physics, forward and backward these two sections motion sequences take place to be used for the linking incident, the steps include: that (1) carry out interpolation with original series and the sequence that searches, obtain joint attitude angle θ after two sections sequence interpolation before and after redirect Blend(t), (2) in the process of physical simulation, the interpolating sequence of Dynamics Controller two sections sequences along front and back generates corresponding reaction action; (3) for each joint, Dynamics Controller adopts proportion differential servomechanism calculating effect moment thereon:
τ = k s ( θ blend ( t ) - θ cur ) - k d ( - θ · cur )
K wherein sAnd k dBe respectively the rigidity and the damping gain of corresponding joint, θ CurWith
Figure A20061005281100112
Represent current joint attitude angle and angular velocity respectively.
5) adopt biomechanical principle that the reaction accompany movement that generates is adjusted, the steps include: that the moment that (1) adopts proportional plus derivative controller to produce will drive the top limbs continuously to the T attitude.When the visual human is subjected to unexpected an impact, will in 0.15 second, be activated based on the controller of biomechanics, adjust visual human's shoulder and elbow joint speed with the skew of reduction centre of body weight.We are foundation with human body in the accident experimental data that slips, and adopt proportional plus derivative controller to calculate the moment τ that drives both arms Rr:
τ rr = k srr ( θ T - pose - θ cur ) - k drr ( - θ · cur )
θ wherein T-poseBe the joint attitude angle under the T shape attitude.
(2) the controller drives both arms are followed the tracks of the touchdown point of health, the action that generates protectiveness is hindered avoiding.The purpose of ACTIVE CONTROL arm is that arm was contacted with ground earlier before shoulder lands, to play the effect of buffering, significant points such as protection head, chest, abdomen.For falling the touchdown point of controller drives both arms tracking health backward and forward; Fall for side direction, controller only drives the touchdown point of the arm tracking health of that side.
(3) adopt the controller drives visual human to arrive or near the initial attitude of motion sequence that will redirect.
This method has been used the method for Parallel Simulation in the physical simulation process.By Parallel Simulation, the reaction action to the unexpected interference in outside can be predicted and calculate to this method.A special physical simulation process is used for predicting simulation result.Therefore, the present invention does not need the results of Physical rewinding, and then creates a new physical simulation process.After obtaining movement locus, just motion sequence that prediction can be come out and the motion sequence in the database compare, and seek suitable redirect sequence and best redirect opportunity, and feed back to normal physical simulation process.
The inventive method no matter be to and the accidental action of environment or the interaction between the visual human, perhaps repeatedly contact all and can obtain effect preferably with the outside.

Claims (6)

1. the reaction accompany movement generation method based on mixing control is characterized in that this method comprises the steps:
1) according to the external event of accepting, the movement locus of prediction visual human under physical simulation, the steps include: that (1) create two identical scenes simultaneously, each scene is under the jurisdiction of a physical simulation process, (2) time step of prognosis modelling process is made as ten times of common simulation, (3) prognosis modelling process and common simulation process are set to start to walk at one time, (4) motion sequence that prediction is come out and the motion sequence in the database compare, and feed back to normal physical simulation process;
2) adopt approximate nearest neighbor search algorithm to carry out motion search, and the method for the level of detail reduction search volume dimension of employing skeletal structure improve the speed of motion search;
3) making similarity between any two frames to motion sequence in movement locus under the physical simulation and the database calculates, difference between sequence that physical simulation obtains and two frames of the sequence in the database, be expressed as the position in corresponding joint and the weights sum of angle difference, establish f sAnd f mRepresent the movement locus of prediction and two frames in the motion sequence in the database respectively, then in a window, two frame difference formula can be expressed as:
Figure A2006100528110002C1
Wherein ws is the size of redirect window, and J is visual human's a joint quantity, p j(f Si) expression f SiThe position in j joint of frame, similarly θ j(f Si) be f SiThe position angle in j joint of frame, T θ, l(f S0, f M0) be a coordinate transform operation, according to the health different parts similarity is calculated the size of contribution, w PjAnd w θ jRespectively to the linear distance and the angular distance weighting in j joint;
4) employing moves the linking incident that forward and backward these two sections motion sequences take place based on the method reaction of formation followability of physics, the steps include: that (1) carry out interpolation with original series and the sequence that searches, obtain joint attitude angle θ after two sections sequence interpolation before and after redirect Blend(t), (2) in the process of physical simulation, the interpolating sequence of Dynamics Controller two sections sequences along front and back generates corresponding reaction action; (3) for each joint, Dynamics Controller adopts proportion differential servomechanism calculating effect moment thereon:
τ = k s ( θ blend ( t ) - θ cur ) - k d ( - θ · cur )
K wherein sAnd k dBe respectively the rigidity and the damping gain of corresponding joint, θ CurWith Represent current joint attitude angle and angular velocity respectively;
5) adopt biomechanical principle that the reaction accompany movement that generates is adjusted; the steps include: that the moment that (1) adopts proportional plus derivative controller to produce will drive the top limbs continuously to the T attitude; (2) the controller drives both arms are followed the tracks of the touchdown point of health, the action that generates protectiveness is hindered avoiding, and (3) adopt the controller drives visual human to arrive or near the initial attitude of motion sequence that will redirect.
2. as claimed in claim 1 a kind of based on the reaction accompany movement generation method of mixing control, it is characterized in that described establishment physical simulation scene is meant: the dynamics human skeleton model of setting up the visual human, be associated with the manikin of hinge arrangement catching the data of getting off, set up hybrid-type visual human's skeleton, the size of each part of visual human's health is determined by the skeleton of catching in the data, the quality of health each several part is the measurement data in the anatomy, and carry out proportional zoom according to the skeleton size of catching in the data, visual human's motion is driven by corresponding motion capture data.
3. as claimed in claim 1 a kind of based on the reaction accompany movement generation method of mixing control, the step that it is characterized in that described Parallel Simulation mechanism is: 1) when because collision or foreign impacts when changing visual human's state, physical simulation process that has big time step is used for apace predicated response result and obtain the track of visual human in this physical simulation process, 2) this is the result of visual human's passive response external event with the motion sequence that the physical simulation process generates fully, be mainly used to search for suitable redirect motion sequence, 3) simultaneously, the motion control that another physical simulation process with step-length normal time will be taken over the visual human, and generation is to the reaction action of external event.
4. as claimed in claim 1 a kind of based on the reaction accompany movement generation method of mixing control, it is characterized in that described motion selection method is: the motion sequence in initial analog track (movement locus that prediction is come out) and the database is compared, rather than as commonsense method, between two sections motion sequences that existed, do similarity and calculate.
5. as claimed in claim 1 a kind of based on the reaction accompany movement generation method of mixing control, it is characterized in that described approximate nearest neighbor search algorithm is: according to the error boundary ε and the data query q of user's appointment, the data p that the space tree search that utilization builds meets the following conditions:
dist(p，q)≤(1+ε)dist(p *，q)
P wherein *Be the real neighbour of q.
6. as claimed in claim 1 a kind of based on the reaction accompany movement generation method of mixing control, the method that it is characterized in that the level of detail reduction search volume dimension of described skeletal structure is: set up the hierarchical model that human body closes skeleton, according to stratum level, reduce the articulation point number successively, and according to search efficiency, by selecting suitable level skeletal structure, on speed and moving-mass, compromise.
CNA2006100528111A 2006-08-07 2006-08-07 Method for generating reaction accompany movement based on hybrid control Pending CN1916969A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CNA2006100528111A CN1916969A (en) 2006-08-07 2006-08-07 Method for generating reaction accompany movement based on hybrid control

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CNA2006100528111A CN1916969A (en) 2006-08-07 2006-08-07 Method for generating reaction accompany movement based on hybrid control

Publications (1)

Publication Number Publication Date
CN1916969A true CN1916969A (en) 2007-02-21

Family

ID=37737959

Family Applications (1)

Application Number Title Priority Date Filing Date
CNA2006100528111A Pending CN1916969A (en) 2006-08-07 2006-08-07 Method for generating reaction accompany movement based on hybrid control

Country Status (1)

Country Link
CN (1) CN1916969A (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101241601B (en) * 2008-02-19 2010-06-02 深圳先进技术研究院 Graphic processing joint center parameter estimation method
CN101840586A (en) * 2010-04-02 2010-09-22 中国科学院计算技术研究所 Method and system for planning motion of virtual human
CN101615301B (en) * 2009-07-29 2013-03-27 腾讯科技（深圳）有限公司 Path control method and system for target in computer virtual environment
CN104732572A (en) * 2014-12-24 2015-06-24 天脉聚源（北京）科技有限公司 Method and device for generating motion image
CN111223168A (en) * 2020-01-17 2020-06-02 腾讯科技（深圳）有限公司 Target object control method and device, storage medium and computer equipment
CN112559308A (en) * 2020-12-11 2021-03-26 广东电力通信科技有限公司 Statistical model-based root alarm analysis method
CN113129415A (en) * 2020-06-22 2021-07-16 完美世界(北京)软件科技发展有限公司 Animation mixed space subdivision method, device, equipment and readable medium

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101241601B (en) * 2008-02-19 2010-06-02 深圳先进技术研究院 Graphic processing joint center parameter estimation method
CN101615301B (en) * 2009-07-29 2013-03-27 腾讯科技（深圳）有限公司 Path control method and system for target in computer virtual environment
CN101840586A (en) * 2010-04-02 2010-09-22 中国科学院计算技术研究所 Method and system for planning motion of virtual human
CN104732572A (en) * 2014-12-24 2015-06-24 天脉聚源（北京）科技有限公司 Method and device for generating motion image
CN104732572B (en) * 2014-12-24 2018-06-19 天脉聚源（北京）科技有限公司 A kind of method and device for generating moving image
CN111223168A (en) * 2020-01-17 2020-06-02 腾讯科技（深圳）有限公司 Target object control method and device, storage medium and computer equipment
CN113129415A (en) * 2020-06-22 2021-07-16 完美世界(北京)软件科技发展有限公司 Animation mixed space subdivision method, device, equipment and readable medium
CN113129416A (en) * 2020-06-22 2021-07-16 完美世界(北京)软件科技发展有限公司 Animation mixed space subdivision method, device, equipment and readable medium
CN112559308A (en) * 2020-12-11 2021-03-26 广东电力通信科技有限公司 Statistical model-based root alarm analysis method
CN112559308B (en) * 2020-12-11 2023-02-28 广东电力通信科技有限公司 Statistical model-based root alarm analysis method

Similar Documents

Publication Publication Date Title
CN1916969A (en) Method for generating reaction accompany movement based on hybrid control
Chen et al. Stabilization approaches for reinforcement learning-based end-to-end autonomous driving
Zordan et al. Motion capture-driven simulations that hit and react
CN104808590B (en) Mobile robot visual servo control method based on key frame strategy
CN101610877B (en) Method and apparatus for haptic control
Billard et al. Discovering optimal imitation strategies
WO2019076044A1 (en) Mobile robot local motion planning method and apparatus and computer storage medium
JP4848515B2 (en) Avatar motion control system, program and method thereof
CN101075352A (en) Laminated barrier-avoiding method for dynamic body
CN103440037A (en) Real-time interaction virtual human body motion control method based on limited input information
CN104299245A (en) Augmented reality tracking method based on neural network
CN112270226A (en) Pedestrian trajectory prediction method based on multi-feature extraction and multi-attention mechanism
CN101515373A (en) Sports interactive animation producing method
CN113276120A (en) Control method and device for mechanical arm movement and computer equipment
CN107578461A (en) A kind of three-dimensional virtual human body physical motion generation method based on subspace screening
CN103729879A (en) Virtual hand stable grabbing method based on force sense calculation
CN104331080B (en) For the fixed point track path planing method of mobile robot
Hu et al. Skiing Simulation Based on Skill‐Guided Motion Planning
CN115018962A (en) Human motion attitude data set generation method based on virtual character model
US20180043245A1 (en) Simulation System for Balance Control in Interactive Motion
CN101952879A (en) Methods and apparatus for designing animatronics units from articulated computer generated characters
CN108724187A (en) A kind of motion planning method and system of robot crawl flyer
Tang et al. Interactive generation of falling motions
CN110046457B (en) Human body model control method and device, electronic equipment and storage medium
Guo et al. Locomotion skills for insects with sample‐based controller

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C12 Rejection of a patent application after its publication
RJ01 Rejection of invention patent application after publication