Deep reinforcement learning network system

A technology of reinforcement learning and network systems, applied in the field of artificial intelligence and box packing, can solve problems such as difficult GPU parallelization, difficult migration, and huge impact, so as to reduce the difficulty of learning and searching, solve the problem of long training time, and reduce the action space Effect

Pending Publication Date: 2021-06-01
SHENZHEN DORABOT ROBOTICS CO LTD
View PDF0 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007]The main disadvantages of the first existing bin packing algorithm based on deep reinforcement learning are: (1) It uses the traditional heuristic algorithm to calculate the box loading position and uses The space occupancy of this loading result is used as a reward for reinforcement learning, so the selected heuristic algorithm will become the bottleneck of the entire boxing algorithm
Since different heuristic algorithms are only optimized for specific datasets, it is difficult for this method to transfer smoothly under different binning scenarios
(2) It is difficult to use GPU parallelization with traditional heuristic algorithms, so that the training process of deep reinforcement learning requires a lot of time and computing resources
[0008] The main disadvantages of the second type of method are: (1) The actual box packing problem usually does not have a fixed loading order, so the scope of application of this type of method is limited
(2) The loading order of the boxes in the box packing problem has a great influence on the final loading result, so it is often difficult for this type of algorithm to optimize the loading rate of the final algorithm
(3) At the same time, since the box loading in the box packing algorithm has strict space constraints, a large amount of calculations are required

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
  • Deep reinforcement learning network system
  • Deep reinforcement learning network system
  • Deep reinforcement learning network system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0036] Preferred embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although preferred embodiments of the present application are shown in the drawings, it should be understood that the present application may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this application will be thorough and complete, and will fully convey the scope of this application to those skilled in the art.

[0037] The terminology used in this application is for the purpose of describing particular embodiments only, and is not intended to limit the application. As used in this application and the appended claims, the singular forms "a", "the", and "the" are intended to include the plural forms as well, unless the context clearly dictates otherwise. It should also be understood that the term "and / or" as used herein refers to and includes a...

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 discloses a deep reinforcement learning network system, which comprises an embedding layer, a sorting layer and a loading layer. The embedding layer is used for mapping packing information of to-be-packed boxes into feature vectors, the sorting layer is used for sorting a plurality of disordered to-be-packed boxes, and the loading layer is used for loading the to-be-packed boxes. The loading layer is used for determining the loading positions and orientations of the to-be-loaded boxes according to the loading sequence, output by the sorting layer, of the to-be-loaded boxes, so the boxing problem is decomposed into two steps, and a layered boxing model is used for solving: 1, a sequence decision problem: giving a series of disordered to-be-loaded boxes, and determining the loading sequence of the to-be-loaded box bodies by using deep reinforcement learning. 2, a loading problem: based on a box loading sequence given by a second algorithm, determining the loading position and orientation of the box by using deep reinforcement learning; wherein the encasement problem is decomposed, so that the whole action space, needing to be searched, of the problem is greatly reduced, and the learning and searching difficulty of the intelligent agent is greatly reduced.

Description

technical field [0001] The present application relates to the field of artificial intelligence and box packing technology, and in particular to a deep reinforcement learning network system and a storage medium. Background technique [0002] At present, boxing is usually carried out manually based on experience. In order to improve the rationality and efficiency of packing, intelligent algorithms can be used to assist packing. [0003] The bin packing problem is a classic academic problem and has a wide range of commercial applications. In the field of logistics, it is often necessary to load a specified series of goods into a specified container, such as a carriage, for transportation. [0004] The vast majority of existing algorithms can be divided into two categories, (1) heuristic search algorithms using artificially formulated rules; (2) optimization algorithms such as genetic algorithms and deep learning algorithms that treat them as nonlinear optimization to solve. ...

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 & AuthorityApplications(China)
IPC IPC(8): G06N3/04G06N3/08G06Q10/08
CPCG06N3/08G06Q10/083G06N3/047G06N3/045
Inventor张经纬资斌
OwnerSHENZHEN DORABOT ROBOTICS CO LTD