System and method for parameter optimization of intelligent robot applying neural network

An intelligent robot and neural network technology, applied in the field of intelligent robot parameter optimization and intelligent robot, can solve the problem of high cost, reduce the dependence, reduce the amount of repeated calculation and calculation

Active Publication Date: 2016-10-12
SHENZHEN FLYING FAIRY INTELLIGENT TECH CO LTD
View PDF4 Cites 16 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0009] For example: using the AlexNet image library to train a neural network, it takes 150 hours to use dual-socket Xeonprocessors (3TFLOPS of computing power), and it takes 2 hours even to use an expensive GPU server dedicated to machine learning, so directly on the terminal robot The cost of implementing deep neural network training on

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • System and method for parameter optimization of intelligent robot applying neural network
  • System and method for parameter optimization of intelligent robot applying neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0057] like figure 2 As shown, the system includes at least one cloud server 2 and at least one terminal robot 1, the terminal robot 1 is connected to the cloud server 2 through a network, and the terminal robot 1 is embedded with a terminal deep neural network module for forward transmission calculation. The server 2 is embedded with a cloud deep neural network module for optimizing parameters according to the sample library.

[0058] The sample library on which the parameter is optimized can be from the terminal robot, or from the storage device or network data of the server.

[0059] Terminal robot 1 includes one or more of the following modules:

[0060] A speech recognition module 4 for recognizing speech environment information;

[0061] A visual recognition module 5 for recognizing three-dimensional environmental information;

[0062] Navigation module 6 for identifying direction environment information;

[0063] A route planning module 7 for performing route plann...

Embodiment 2

[0076] This embodiment provides a method for optimizing parameters of an intelligent robot using a neural network, which is further described by taking a deep neural network as an example. The method for parameter optimization includes the following steps:

[0077] Step 1, the terminal robot 1 recognizes and collects environmental information and action instructions;

[0078] Step 2, the terminal deep neural network module performs forward transmission calculation on different types of information;

[0079] Step 3. The cloud deep neural network module optimizes and stores parameters of different types of information according to the sample library;

[0080] Step 4. Determine whether to compress the optimized parameters. If yes, execute step 5. If not, terminate the execution and issue a system prompt;

[0081] Step 5: Compress the optimized parameters, and send the compressed parameters to the terminal robot 1 .

[0082] The environment information includes voice environment...

Embodiment 3

[0091] This embodiment is described by taking the application of the present invention to parameter optimization and compression of visual recognition information as an example.

[0092] (1) A neural network module for realizing visual recognition information is embedded on the cloud server;

[0093] (2) The cloud server collects the marked image data obtained from the database, the network or from the terminal robot in the system, and uses it as the input of the node parameters for training the neural network, and realizes the optimization calculation of the parameters through specific operations;

[0094] (3) After the cloud server realizes the calculation of parameters, according to the actual network parameter results and the computing power of the target terminal robot, you can choose whether to optimize the forward calculation of the entire network, including reducing the parameter accuracy and the number of nodes and connections in the network , and you can also choose ...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention relates to a system and a method for parameter optimization of an intelligent robot applying a neural network and relates to the field of parameter optimization of the intelligent robot. Parameter optimization and compression processes are completed through a cloud server, so that the calculated amount of a terminal robot is reduced. The system comprises at least one cloud server and at least one terminal robot, wherein the terminal robot is connected to the cloud server through a network; a terminal neural network module used for forward transmission and calculation is embedded into the terminal robot; and a cloud neural network module used for parameter optimization according to a sample library is embedded into the cloud server. The method comprises the following steps that the terminal robot carries out identification and collection on environment information and an action command; different cloud neural network modules carry out parameter optimization and storage on different kinds of information; the terminal robot judges whether to compress the parameter optimization process of each neural network module; and the terminal robot compresses the optimized parameters. The system and the method are suitable for parameter optimization of each neural network module of the intelligent robot.

Description

technical field [0001] The invention relates to the field of intelligent robots, in particular to the field of parameter optimization of intelligent robots. Background technique [0002] At present, most of the existing intelligent robots are stand-alone versions, that is, they do not achieve better performance through remote servers. These robots have poor hardware conditions, high hardware costs, and low levels of intelligence that can be realized. Bad experience. [0003] In order to overcome the shortcomings of the stand-alone version of the robot, more and more cloud robots are gradually appearing, and the cloud robot is served by the server in the cloud of the application. [0004] The title of the Chinese patent invention is "Cloud Robot System and Implementation Method", Publication No.: 101973031A, Publication Date: 2011-02-16, using a remote cloud computing platform to directly send control commands through the network. [0005] The title of the Chinese patent in...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): H04L12/24G06N3/04
CPCH04L41/0246H04L41/0803H04L41/0813G06N3/045
Inventor 简卫朱奇良
Owner SHENZHEN FLYING FAIRY INTELLIGENT TECH CO LTD
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap