Technical Analysis and Expectation for Cotton Harvesting Based on Agricultural Robot
Changying Ji
- Year
- 2006
- Citations
- 3
Abstract
Differences of cotton between America and China exist in harvesting and classification were summarized.In America,the characteristics of cotton production have been recognized,such as large scale,single variety,long autumn and mid-long fibre etc.It helps to pick and classify cotton fully by machines.In contrast,in China,the characteristics of cotton production leads to inadaptation for mechanism-based technology.Cotton harvesting mostly performed by hand,mainly restricted by plenty of labor,and the classification that is fulfilled by dint of human sensing.In view of the conjuncture,an assumption for cotton harvesting based on agricultural robot were put forward in this paper.In China,agricultural robot-based cotton harvesting combines the advantages of the handcraft-based and the mechanism-based,and it has the potential to resolve the conflict between production efficiency and high fiber grade.Comparing with mechanism-harvested cotton,robot-harvested cotton includes the follow advantages:⑴Adapting the variety of cotton.Many qualification of mechanism-based technology for cotton harvesting such as mid-long fibre,short-term autumn,single variety,and poor disease resistance,result in the use of agricultural robot technology for cotton harvesting.Owing to human simulating picker,robot-based technology with programmable feature resolves some problems such as short fibre,long-term autumn,variety because of the special climate and zone.⑵Requesting poor agriculture.In order to control the height along with maturity of cotton plant,and reduce the impurity mixed with leaves,the use of chemicals for pre-harvest cotton will lead to drug-resistance before cotton harvesting.Agriculture robot with spatial freedom has the potential to avoid leaf scrap mixed,reduce cost and ensure sowing of cotton plant.⑶Increasing the yield of high quality cotton.Traditional visual quality inspection performed by human inspectors has the potential to be replaced by computer vision systems for robot-harvested cotton,and a nicer representative sample may meet the classification before cotton harvesting.The quality of classification for cotton harvesting as well as the yield of high quality cotton will be advanced greatly.⑷Reducing production cost.Agricultural robot-harvested ensures high quality and avoids plenty of fee for picking owe to simulating human being's action.harvesting robot plays some other roles with compact,flexible moving and convenient for remote portage in agricultural production.Hence,cotton harvesting using agriculture robot adapts to the situation of China.At present,harvest robot with machine vision has been capable of harvesting many fruit and vegetable such as orange,tangerine,tomato,murphy,fungus,cucumber,apple,watermelon,grape.Plenty of experience obtained in terms of harvesting consists of three aspects:⑴robot's self-determination navigation,⑵objective detection,⑶robot's framework.Digital image processing performed with a computer to manipulate information within an image makes it useful.Image processing in agricultural applications may consist of three steps: ⑴ image enhancement,⑵image feature extraction,and ⑶image feature classification.The image acquisition board receives imaging data from a CCD camera.Image enhancement procedures such as pixel-to-pixel operations,filters,and morphological operations are generally used to correct some problems such as poor contrast or noise caused by inadequate and non-uniform illumination.Statistical procedures such as mean,standard deviation,variance and principle component analysis can be used to extract cotton features,such as size,shape,shine,color,texture,position,leaf states of cotton.Modern mathematics microscope,wavelet analysis system,can be successfully applied to edge detection due to analyzing locally in time.For the rapid prototype of a machine vision system,artificial intelligence programming can be incorporated into the system.Newer tools such as artificial neural network-Fuzzy/Roug
Keywords
Related papers
Statistical Learning Theory
Yuhai Wu, Vladimir Vapnik
1999
Artificial intelligence: a modern approach
1995
Applied Nonlinear Control
Jean-Jacques Slotine, Weiping Li
1991
A new optimizer using particle swarm theory
R.C. Eberhart, James Kennedy
2002