Space and time-division double-flow data-driven deep Q learning method for autonomous intelligent navigation control of unmanned ship
A control method, unmanned ship technology, applied in two-dimensional position/channel control, adaptive control, general control system and other directions, can solve the problem of not using artificial intelligence methods, etc., and meet the requirements of good compatibility and software and hardware resources Simple, highly artificial intelligence
Inactive Publication Date: 2019-05-28
SHANGHAI MARITIME UNIVERSITY
View PDF10 Cites 8 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
The patent application number is 201810008481.9, which invented a cooperative cloud control system for autonomous navigation of unmanned ships, but it requires the common interaction and function of very complex systems and information such as shore, shipboard, communication and collaborative cloud control systems
Patents with application numbers 201710691295.5, 201711285895.8 and 201810160232.1 invented an autonomous navigation system and method for unmanned ships, but none of them used artificial intelligence methods
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 moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0041] The present invention will be further elaborated below by describing a preferred specific embodiment in detail in conjunction with the accompanying drawings.
[0042] 1. Sampling space-time dual-stream big data
[0043] Through the 360° pulse laser rangefinder installed on the top of the unmanned ship 1, the distance d between the unmanned ship 1 and the surrounding environment 2 is scanned with an angular resolution of 1.8° t space big data, that is, to measure the distance d between the unmanned ship 1 and the surrounding environment 2 of 200 dimensions per frame t spatial big data. Then pass through two adjacent frames d t difference in data o t = d t -d t-1 , to measure the relative motion speed o of the 200-dimensional unmanned ship 1 and the surrounding environment 2 per frame t The time-division big data of , where the subscript t represents the sampling time t. Such as figure 1 360° pulsed laser rangefinder data in o t , d t 4.
[0044] 2. Design deep...
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
Login to View More Abstract
The invention provides a space and time-division double-flow large-data-driven deep Q learning network method for achieving autonomous intelligent navigation control of an unmanned ship under high-precision navigation, which comprises the following specific steps: sampling space-time double-flow large data; designing a deep Q learning network intelligent obstacle avoidance controller; designing areward and punishment function; designing an intelligent switching threshold function; carrying out on-line learning. The method can achieve that the unmanned ship navigates under high-precision positioning navigation when in open water areas; the depth Q learning network intelligent obstacle avoidance controller enables the unmanned ship to navigate independently in an intelligent obstacle avoidance mode when in complex water areas; real-time risk estimation factors can be sampled and evaluated according to the environment, so that real-time intelligent switching between the two modes is achieved. In addition, the deep Q learning network intelligent obstacle avoidance controller has self-learning capability and high artificial intelligence. Finally, the method has better compatibility tothe existing ship navigation control system, and the requirement of software and hardware resources for achieving the method is relatively simple.
Description
technical field [0001] The invention relates to an intelligent navigation control method for an unmanned ship driven by deep Q-learning with space-time dual-stream data, in particular to an unmanned ship based on a deep Q-learning network driven by space and time-division dual-stream real-time sampling data under high-precision positioning and navigation. A ship intelligent navigation control method. The invention belongs to the technical field of unmanned ship intelligent control. Background technique [0002] It is not an easy task for ships under high-precision positioning and navigation to have human observation capabilities and intelligence, without relying on the driver's lookout and steering, and to achieve autonomous intelligent navigation and obstacle avoidance through complex water surfaces. Due to the open water surface and variable obstacle positions, unmanned ships cannot rely on lane line detection like unmanned vehicles; nor can they perform 3D modeling like ...
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
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G05D1/02G05B13/02G05B13/04G06N3/04
Inventor 黄志坚随博文温家一吴恭兴张桂臣刘雁集
Owner SHANGHAI MARITIME UNIVERSITY