CN115543087B - Artificial intelligence scoring method for virtual environment skill practice - Google Patents

Artificial intelligence scoring method for virtual environment skill practice Download PDF

Info

Publication number
CN115543087B
CN115543087B CN202211256492.1A CN202211256492A CN115543087B CN 115543087 B CN115543087 B CN 115543087B CN 202211256492 A CN202211256492 A CN 202211256492A CN 115543087 B CN115543087 B CN 115543087B
Authority
CN
China
Prior art keywords
algorithm
skill
scoring
action
model
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.)
Active
Application number
CN202211256492.1A
Other languages
Chinese (zh)
Other versions
CN115543087A (en
Inventor
李力
韩旭
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.)
Guangzhou Qiangji Information Technology Co ltd
Original Assignee
Guangzhou Qiangji Information Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Guangzhou Qiangji Information Technology Co ltd filed Critical Guangzhou Qiangji Information Technology Co ltd
Priority to CN202211256492.1A priority Critical patent/CN115543087B/en
Publication of CN115543087A publication Critical patent/CN115543087A/en
Application granted granted Critical
Publication of CN115543087B publication Critical patent/CN115543087B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
    • G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
    • G—PHYSICS
    • G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
    • G09B—EDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
    • G09B19/00—Teaching not covered by other main groups of this subclass
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Educational Administration (AREA)
  • Educational Technology (AREA)
  • Human Computer Interaction (AREA)
  • Electrically Operated Instructional Devices (AREA)

Abstract

The invention provides an artificial intelligence scoring method for virtual environment skill practice, which comprises a model training stage, wherein the model training stage comprises the steps of collecting and recording skill action data of N technicians through an AI system, learning standard actions of the N technicians and adjusting standard parameters of each skill action; calculating the difference value between the input action and the standard action by using ten algorithms, adjusting the calculation weights of the ten distance algorithms by using a three-layer AI neural network algorithm, and scoring the average value weighted by each algorithm; wearing the VR equipment by the master of the art and moving, gathering master of the art and moving data through the VR system, and recording the state change of the operated object after the skill action input of the master of the art. The artificial intelligence scoring method for virtual environment skill practice can realize real-time calculation and scoring of action results, solve the problem of dependence of a mechanic on the environment, establish a standard objective skill scoring standard, and improve the virtual environment skill practice of skills.

Description

Artificial intelligence scoring method for virtual environment skill practice
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to an artificial intelligence scoring method for virtual environment skill practice.
Background
The existing skill practice score has the following defects:
(1)skilled workerThe training or examination of (1) needs practice operations with great weight, and the practice operations are environmentally dependent on practice objects in real scenes, such as: the operation of the infant nurses is to nurse the infants, and the actual practice training and evaluation cannot be carried out by the real infants under the actual condition, so that the actual practice scoring is difficult.
(2) The existing practice depends heavily on subjective judgment when taking examination scores, and the judgment standard is different from person to person, so that the scores are unfair.
(3) At present, the number of the current day,skilled workerThe application of the real operation in the virtual reality is basically in one-way output of teaching, the virtual reality can only realize interaction between the limbs with large movement and the virtual environment, and for fine real operation movement, the result form after the operation is difficult to be accurately reflected in a virtual scene.
Disclosure of Invention
The invention provides a method for realizing real-time calculation and scoring of action resultsSkilled workerThe dependence of actual operation on environment is establishedObjective practice scoring criteria such thatSkilled workerThe artificial intelligence scoring method for virtual environment skill practice can sense the state change of each step of operation to the object to be operated in real time during the actual operation, thereby improving the skill.
The technical scheme adopted by the invention is as follows: an artificial intelligence scoring method for virtual environment skill practice comprises the following steps: a model training phase comprising the steps of:
and (3) finishing an AI scoring model:
n number ofMaster skilled in the artWearing motion capture equipment or optical dynamic capture equipment, collecting and recording skill motion data through an AI system, and gathering all motion data to form an N-dimensional array; by aligning NMaster skilled in the artThe standard parameters of each skill action are adjusted to obtain an N-dimensional standard array;
respectively calculating the difference value of the input action and the standard action by using a Manhattan distance algorithm, a Chebyshev distance algorithm, a Minkowski distance algorithm, a standardized Euclidean distance algorithm, a Mahalanobis distance algorithm, an included angle cosine algorithm, a Hamming distance algorithm, a Jacard distance algorithm, a correlation coefficient algorithm and an information entropy algorithm, and simultaneously adopting the Manhattan distance algorithm, the Chebyshev distance algorithm, the Minkowski distance algorithm, the normalized Euclidean distance algorithm and the information entropy algorithmThree-layer AI neural network algorithmAdjusting the calculation weights of the ten distance algorithms, and taking the weighted average value of the algorithms for scoring, wherein the scoring algorithm is as follows:
Figure BDA0003889780060000021
wherein n is the number of examination actions of skill practice, w j For the weight of each distance algorithm, d ij Difference values from standard movements under different distance algorithms for each movement;
if each oneMaster skilled in the artIf the score of the standard action is more than 95 points, outputting an AI score model; if the score is less than or equal to 95 points, the weights of various distance algorithms are adjusted in a reverse derivation mode, and the process is repeated until all standard actions are scoredMore than 95 minutes;
and (3) completing parameter fitting of the VR model:
byMaster skilled in the artWearing VR equipment to move, and collecting through VR systemMaster skilled in the artMotion data of the operated objectMaster skilled in the artThe state change after the skill action input of (1),master skilled in the artAnd judging whether the model changes reasonably according to experience, if not, adjusting the parameters of the operated object, so that the influence of each action on the operated object accords with an objective rule, and outputting a VR model.
Further, the scoring method also comprises a model application stage,
master skilled in the artWhen practical exercise or examination is conducted in the application stage, dynamic capture equipment and VR equipment are deployed, skill operation is conducted according to VR prompts, and the dynamic capture equipment collectsMaster skilled in the artThe action information is respectively recorded into a VR model and an AI scoring model and provided by the VR modelMaster skilled in the artAnd displaying the latest state of the operated object, guiding the next operation, and giving a real-time score by the AI scoring model.
Further, theThree-layer AI neural network algorithmComprises an input layer, an output layer and a hidden layer; the number of the input layers and the number of the output layers are consistent with the number of the distance algorithms, and the number of the nodes of the hidden layer is larger than or equal to the number of the distance algorithms.
Compared with the prior art, the artificial intelligence scoring method for virtual environment skill practice provided by the inventionMaster skilled in the artUnder the support of the dynamic capture equipment, the motion data is sent to an artificial intelligence system, and the real-time calculation and scoring of motion results are realized; solve the problem ofMaster skilled in the artThe dependence of the real exercises on the environment establishes a standard objective real exercise scoring standard, so thatSkilled engineer Friedel-craftsWhen the real operation is carried out, the state change of the object to be operated by each step of operation can be sensed in real time, so that the skill can be learned and improved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description serve to explain the principles of the invention and not to limit the invention. In the drawings, there is shown in the drawings,
FIG. 1: the invention discloses a skill action scoring flow chart;
FIG. 2: the weight adjustment neural network schematic diagram of ten algorithms is provided;
FIG. 3: the invention relates to an AI algorithm training flow chart of practice action scoring;
FIG. 4: the invention relates to a virtual reality simulation parameter correction chart;
FIG. 5: application phase of the inventionMaster skilled in the artAnd (4) an actual operation scoring flow chart.
Detailed Description
The following detailed description of embodiments of the invention refers to the accompanying drawings. It should be understood that the detailed description and specific examples, while indicating the present invention, are given by way of illustration and explanation only, not limitation.
As shown in fig. 1 to 3, the artificial intelligence scoring method for virtual environment skill practice of the present invention includes a model training phase and a model application phase; in the model training stage, parameter fitting between an AI (Artificial Intelligence) scoring model and a VR (Virtual Reality) model needs to be completed.
(I) complete AI scoring model
First, NMaster skilled in the artWearing a motion capture device or an optical dynamic capture device, collecting and recording data of skill motion through an AI system (such as an artificial intelligence edge computing box), and gathering all motion data to form an N-dimensional array as follows:
input in = [ x ] 1 ，x 2 ，x 3 ……x n ]；
By aligning NMaster skilled in the artThe standard parameters of each skill action are adjusted to obtain an N-dimensional standard array as follows:
standard std = [ y = 1 ，y 2 ，y 3 ……y n ]。
Then applying a Manhattan distance algorithm, a Chebyshev distance algorithm, a Minkowski distance algorithm, a normalized Euclidean distance algorithm,The Mahalanobis distance algorithm, the included angle cosine algorithm, the Hamming distance algorithm, the Jacard distance algorithm, the correlation coefficient algorithm and the information entropy algorithm are adopted to respectively calculate the difference value of the input action and the standard action, and meanwhile, the difference value is calculated by adopting the Markov distance algorithm, the included angle cosine algorithm, the Hamming distance algorithm, the Jacard distance algorithm, the correlation coefficient algorithm and the information entropy algorithmThree-layer AI neural network AlgorithmAdjusting the calculation weights of the ten distance algorithms, and taking the weighted average values of the algorithms for scoring;
assuming that n assessment actions of a skill real exercise are adopted, the ten distance algorithms are adopted, and the weight of each distance algorithm is w j The difference between each action and the standard action is d under different distance algorithms ij The scoring algorithm is then as follows:
Figure BDA0003889780060000041
if each oneMaster skilled in the artIf the score of the standard action is more than 95 points, outputting an AI score model; and if the score is less than or equal to 95 points, adjusting the weights of the network nodes of various distance algorithms in a reverse derivation mode, and repeating the process until the scores of all the standard actions are greater than 95 points.
In this step, taking a normalized euclidean distance algorithm as an example, the calculation difference formula is as follows:
Figure BDA0003889780060000042
the remaining nine distance algorithms calculate differences according to their respective standard formulas.
Referring to FIG. 3, weights for ten distance algorithms are calculatedThree-layer AI neural network algorithmNamely: the number of the input layers and the output layers is consistent with the number of the distance algorithms, and if the number of the distance algorithms is n (n =10 in the invention), the number of the nodes of the hidden layer can be adjusted according to requirements, and the requirement is greater than or equal to the number of the distance algorithms (namely ≧ n).
Weight w in Manhattan distance algorithm 1 For example, the calculation process is as follows
x 1 ＝d 1
Figure BDA0003889780060000043
Figure BDA0003889780060000044
w 1 ＝z 1
Assuming that the final system yields a score of 80, expected to be 100, then the algorithm error is-20 and the weights for the manhattan distances are adjusted as follows:
let Δ w 1 Error value to be adjusted, w 1 ' As adjusted value, assume learning rate as eta, eta<1, the calculation result in the case of an error of d (d = 20) is as follows:
Figure BDA0003889780060000045
w 1 ′＝w 1 +Δw 1
let Δ w 12 Error value to be adjusted, w 12 ' is the adjusted value, the calculation result is as follows:
Figure BDA0003889780060000051
w 12 ′＝w 12 +Δw 12
let Δ w 11 Error value to be adjusted, w 11 ' is the adjusted value, the calculation result is as follows:
Figure BDA0003889780060000052
w 11 ′＝w 11 +Δw 11
therefore, a training process is completed, the weight is adjusted and optimized, and the process is repeated repeatedly to adjust the weight value so as to achieve the purpose of correct scoring.
The weight calculation process of the remaining nine distance algorithms is the same as the above calculation process, and is not described herein again.
And (II) completing parameter fitting of the VR model:
as shown in FIG. 4, is composed ofMaster skilled in the artWearing VR equipment to move, and collecting through VR systemMaster skilled in the artMotion data, recording the operated object (i.e. training VR model)Master skilled in the artThe state change after the skill action input of (1),master skilled in the artAnd judging whether the change of the operated object is reasonable or not according to experience, if not, adjusting the parameters of the operated object, so that the influence of each action on the operated object accords with an objective rule, and outputting a VR model.
(III) model application phase
As shown in fig. 5, finally, entering a model application stage, the parameters are not adjusted any more, and if the scores are not reasonable, returning to the model training stage again;master skilled in the artWhen practical exercise or examination is conducted in the application stage, dynamic capture equipment and VR equipment are deployed, skill operation is conducted according to VR prompts, and the dynamic capture equipment collectsMaster skilled in the artRespectively inputting the action information into a VR model and an AI scoring model, and providing the action information by the VR modelMaster skilled in the artAnd displaying the latest state of the operated object so as to guide the next operation, wherein the AI scoring model gives a real-time score.
In conclusion, the artificial intelligence scoring method for virtual environment skill practice of the invention utilizes the motion capture technology to collectExercise machine Master workerStandard skill action, finishing model training of standard action through AI, and simultaneously carrying out standard parameter calibration on VR model to ensure thatMaster skilled in the artThe skill actions are accurately acquired through the action capturing equipment in the virtual reality, and the model state after the operation is updated in real time in the VR model while the skill actions are scored in real time in the AI model, so that the effect of realizing the real-time skill actionMaster skilled in the artTraining and examination of the skill of practice can be completed in the virtual reality environment; so thatMaster skilled in the artAt no dependence onMaster skilled in the artThe teachers can still learn by themselves under the condition of the operating environmentAnd (5) skill training. Furthermore, by using a scoring system for AI, dependence can be madeMaster skilled in the artThe practical operation link of the master subjective scoring realizes the objectification and standardization.
Any combination of the various embodiments of the present invention should be considered as disclosed in the present invention, unless the inventive concept is contrary to the present invention; within the scope of the technical idea of the invention, any combination of various simple modifications and different embodiments of the technical solution without departing from the inventive idea of the present invention shall fall within the protection scope of the present invention.

Claims (3)

1. An artificial intelligence scoring method for virtual environment skill practice is characterized by comprising the following steps: a model training phase comprising the steps of:
and (3) finishing an AI scoring model:
the method comprises the following steps that N technicians wear motion capture equipment or optical dynamic capture equipment, data of skill motions are collected and recorded through an AI system, and all motion data are gathered to form an N-dimensional array; through the learning of the standard actions of N technicians, the standard parameters of each skill action are adjusted to obtain an N-dimensional standard array;
calculating the difference value of the input action and the standard action by using a Manhattan distance algorithm, a Chebyshev distance algorithm, a Minkowski distance algorithm, a standardized Euclidean distance algorithm, a Mahalanobis distance algorithm, an included angle cosine algorithm, a Hamming distance algorithm, a Jacard distance algorithm, a correlation coefficient algorithm and an information entropy algorithm, simultaneously adjusting the calculation weights of the ten distance algorithms by using a three-layer AI neural network algorithm, and taking the weighted average value of the algorithms for scoring, wherein the scoring algorithm is as follows:
Figure FDA0004111893340000011
wherein n is the number of examination actions of skill practice, w j For the weight of each distance algorithm, d ij For each actionDifference from standard action under different distance algorithms;
if the score of the standard action of each technician master is greater than 95 points, an AI score model is output; if the score is less than or equal to 95 points, adjusting the weights of various distance algorithms in a reverse derivation mode, and repeating the process until the scores of all standard actions are greater than 95 points;
and (3) completing parameter fitting of the VR model:
wearing the VR equipment by the master of the art and moving, gathering master of the art and moving data through the VR system, recording by the state change of operation thing after master of the art and moving input of the master of the art and moving, whether the master of the art and moving is reasonable according to the experience judgment model, unreasonable then adjust the parameter of the operated thing for every action accords with objective law to the influence of the operated thing, outputs the VR model.
2. The method for artificial intelligence scoring of skill practices in a virtual environment of claim 1, wherein: the scoring method further comprises a model application phase,
when a technician does actual practice or examination in an application stage, dynamic capture equipment and VR equipment are deployed, skill operation is carried out according to VR prompts, the dynamic capture equipment collects action information of the technician and respectively records a VR model and an AI scoring model, the VR model provides latest state display of an operated object operated by the technician, next-step operation is guided, and the AI scoring model gives real-time scores.
3. The method for artificial intelligence scoring of skill practices in a virtual environment of claim 1, wherein: the three-layer AI neural network algorithm comprises an input layer, an output layer and a hidden layer; the number of the input layers and the number of the output layers are consistent with the number of the distance algorithms, and the number of the nodes of the hidden layer is larger than or equal to the number of the distance algorithms.
CN202211256492.1A 2022-10-14 2022-10-14 Artificial intelligence scoring method for virtual environment skill practice Active CN115543087B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211256492.1A CN115543087B (en) 2022-10-14 2022-10-14 Artificial intelligence scoring method for virtual environment skill practice

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211256492.1A CN115543087B (en) 2022-10-14 2022-10-14 Artificial intelligence scoring method for virtual environment skill practice

Publications (2)

Publication Number Publication Date
CN115543087A CN115543087A (en) 2022-12-30
CN115543087B true CN115543087B (en) 2023-04-07

Family

ID=84733957

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211256492.1A Active CN115543087B (en) 2022-10-14 2022-10-14 Artificial intelligence scoring method for virtual environment skill practice

Country Status (1)

Country Link
CN (1) CN115543087B (en)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108764120A (en) * 2018-05-24 2018-11-06 杭州师范大学 A kind of human body specification action evaluation method
CN112057834A (en) * 2020-09-10 2020-12-11 青岛大学 Rehabilitation action standard judging method based on sensor
CN112365177A (en) * 2020-11-20 2021-02-12 浙江工业大学 Evaluation method of automobile maintenance training based on VR
CN112933581A (en) * 2021-02-02 2021-06-11 江西服装学院 Sports action scoring method and device based on virtual reality technology
CN113867532A (en) * 2021-09-30 2021-12-31 上海千丘智能科技有限公司 Evaluation system and evaluation method based on virtual reality skill training
WO2022060241A1 (en) * 2020-09-18 2022-03-24 Публичное Акционерное Общество "Сбербанк России" Interactive training device for carrying out training with the aid of virtual reality

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108764120A (en) * 2018-05-24 2018-11-06 杭州师范大学 A kind of human body specification action evaluation method
CN112057834A (en) * 2020-09-10 2020-12-11 青岛大学 Rehabilitation action standard judging method based on sensor
WO2022060241A1 (en) * 2020-09-18 2022-03-24 Публичное Акционерное Общество "Сбербанк России" Interactive training device for carrying out training with the aid of virtual reality
CN112365177A (en) * 2020-11-20 2021-02-12 浙江工业大学 Evaluation method of automobile maintenance training based on VR
CN112933581A (en) * 2021-02-02 2021-06-11 江西服装学院 Sports action scoring method and device based on virtual reality technology
CN113867532A (en) * 2021-09-30 2021-12-31 上海千丘智能科技有限公司 Evaluation system and evaluation method based on virtual reality skill training

Also Published As

Publication number Publication date
CN115543087A (en) 2022-12-30

Similar Documents

Publication Publication Date Title
CN110781838B (en) Multi-mode track prediction method for pedestrians in complex scene
CN103473801B (en) A kind of human face expression edit methods based on single camera Yu movement capturing data
CN108776796B (en) Action identification method based on global space-time attention model
CN111079561A (en) Robot intelligent grabbing method based on virtual training
CN103544483B (en) A kind of joint objective method for tracing based on local rarefaction representation and system thereof
CN106373453A (en) Intelligent immersive high-speed train virtual driving behavior evaluation method and simulation system
CN107240049B (en) Automatic evaluation method and system for remote action teaching quality in immersive environment
CN109543285B (en) Crowd evacuation simulation method and system integrating data driving and reinforcement learning
CN115661943B (en) Fall detection method based on lightweight attitude assessment network
CN112434679B (en) Rehabilitation exercise evaluation method and device, equipment and storage medium
CN115484410B (en) Event camera video reconstruction method based on deep learning
CN112507920A (en) Examination abnormal behavior identification method based on time displacement and attention mechanism
CN115775085A (en) Smart city management method and system based on digital twin
CN115543087B (en) Artificial intelligence scoring method for virtual environment skill practice
CN115239508A (en) Scene planning adjustment method, device, equipment and medium based on artificial intelligence
US7486815B2 (en) Method and apparatus for scene learning and three-dimensional tracking using stereo video cameras
CN103793054B (en) A kind of action identification method simulating declarative memory process
Chen Research on college physical education model based on virtual crowd simulation and digital media
CN112991544A (en) Group evacuation behavior simulation method based on panoramic image modeling
CN114504777B (en) Exercise intensity calculation system and method based on neural network and fuzzy comprehensive evaluation
CN114863352B (en) Personnel group behavior monitoring method based on video analysis
CN116186561A (en) Running gesture recognition and correction method and system based on high-dimensional time sequence diagram network
CN110705613A (en) Object classification method
CN105224952A (en) Based on the double interbehavior recognition methods of largest interval markov pessimistic concurrency control
CN115880111A (en) Virtual simulation training classroom teaching management method and system based on images

Legal Events

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