Zhongxiang Zhu

Image Segmentation Based on Ant Colony Optimization and K-Means Clustering

According to the characteristics of the ant colony optimization and the K-means clustering, a method for the image segmentation based on the ant colony optimization and the K-means clustering was proposed in this paper. Firstly, the basic principle of the two algorithms were introduced. Secondly, their characteristics on the image segmentation were analyzed. Finally the improved algorithm was proposed, this algorithm can effectively overcome shortages which are the slow rate of the ant colony optimization and the K-means clustering dependent on the initial clustering centers. Experimental results proved that the improved algorithm was an effective method for the image segmentation in the practical application, which could segment the object accurately.

Research on Traffic Number Recognition Based on Neural Network and Invariant Moments

Traffic number recognition is the important and essential content on license plate recognition and traffic sign recognition. A method of traffic number recognition based on the neural network and the invariant moments was proposed in this paper. Firstly, the area of the traffic number was located from the complicated image background and each number was taken by the image segmentation. Secondly, the features of each number were obtained by Hu invariant moments, which are the invariability of the translation, the ratio and the rotation, and have lower computational complexity. Finally, the traffic number was recognized by the BP neural network. Experimental results proved that the proposed method can be used for fast and efficient recognition of the traffic number with high accuracy.

Study on Crop Image Feature Extraction of Vehicle-Based Estimation System on Large Scale Crop Acreage

Vehicle-based estimation system on large scale crop acreage, which was equipped with GPS receivers, GIS software and a video camera on a off-road vehicle, can capture the images of video and calculate the crop acreage proportion based on matching between GPS information and image recognition. The system provides the credible data for government to make decision and provides technological method. In order to enhance the real-time and reliable character of the system, namely improve the precision of image recognition and estimation result, the grid algorithm of the crop image feature extraction was proposed. The principle of the grid algorithm was that the every crop image was segmented to 16 grids averagely and the four corner grids were regarded as the beginning area during feature extraction and crop recognition. Whether or not to continue extracting feature and recognizing in other grids as well as the recognizing sequence was determined according to the result of crop recognition in the four beginning areas. The texture feature, the shape feature and the color feature were extracted in terms of the particularity of the crop recognition, then the different features or the feature combination were used in order to recognize the crop. The detailed analysis methods of different crops and different cultivating condition were discussed in this paper.

Expert System for Tractor Cab Man-Machine Interface Evaluation

The expert system for tractor cab man-machine interface evaluation was developed using Visual C++ and CLIPS that is one of the expert system tools on the re-development platform of CATIA. The system structure and functions were presented. The knowledge expression, reason mechanism and the design and build method of knowledge base were put forward. The expert system can be used to evaluate the man-machine interface of tractor cab quickly and find the shortage of design. Finally, some improvement suggestions were given by the system. Experiments were carried out for a tractor model. The results showed that the expert system could evaluate the tractor model man-machine interface effectively.

Control System for Tractor-Platooning

A control system for two-tractor platooning was developed, which dealt with follow-up motions for both straight and curvous courses. Firstly, a trajectory of the leading tractor was dynamically obtained by processing the position points of the leading tractor with the method of least squares and curve fitting. Secondly, based on the vehicle kinematic model, a reference course for the following tractor was dynamically created from the trajectory of the leading tractor with a given lateral offset. Finally, an optimal path-tracking controller was designed to guide the following tractor along the reference course. Field tests were conducted on a flat meadow. The test results indicated that the following tractor followed up the leading tractor successfully. The mean and RMS lateral deviations for the straight course were 0.02 and 0.02 m, respectively, whereas those for the sinusoidal course were 0.02 and 0.04 m, respectively.

Vision System Calibration of Agricultural Wheeled-Mobile Robot Based on BP Neural Network

Camera calibration is the base of the machine vision based the autonomous navigation of the agricultural wheeled-mobile robots. There are the complex nonlinear relationship between the actual position points and the matched image points. Therefore the camera parameters have to be calculated by a precise imaging model. The more precise the imaging model requires, the more complicated the calibration becomes. It was proved that some traditional calibration methods, such as the method of Zhengyou Zhang, were inconvenient and their accuracy were also low. In this paper, according to the characteristic of the BP neural network, which can express any nonlinear relationship between inputs and outputs, a new method based on the BP neural network was applied to calibrate the vision system of an agricultural wheeled-mobile robot. The experimental results showed that this method was feasible and accurate.

Proportional directional valve based automatic steering system for tractors

Most automatic steering systems for large tractors are designed with hydraulic systems that run on either constant flow or constant pressure. Such designs are limited in adaptability and applicability. Moreover, their control valves can unload in the neutral position and eventually lead to serious hydraulic leakage over long operation periods. In response to the problems noted above, a multifunctional automatic hydraulic steering circuit is presented. The system design is composed of a 5-way-3- position proportional directional valve, two pilot-controlled check valves, a pressure-compensated directional valve, a pressurecompensated flow regulator valve, a load shuttle valve, and a check valve, among other components. It is adaptable to most open-center systems with constant flow supply and closed-center systems with load feedback. The design maintains the lowest pressure under load feedback and stays at the neutral position during unloading, thus meeting the requirements for steering. The steering controller is based on proportional-integral-derivative (PID) running on a 51-microcontroller-unit master control chip. An experimental platform is developed to establish the basic characteristics of the system subject to stepwise inputs and sinusoidal tracking. Test results show that the system design demonstrates excellent control accuracy, fast response, and negligible leak during long operation periods.

Virtual Reality and the Application in Virtual Experiment for Agricultural Equipment

The meaning and characteristics of the virtual experiment and the significance of applying virtual prototype into agricultural equipment were analyzed in this paper. Then a kind of virtual experiments system platform on agricultural equipments was founded by using MultiGen Creator, Vega Prime software and VC++ programming language. According to the functions of the virtual experiment system, the structure and all components of the system were introduced. Furthermore, the methods of network design, synchronous driving, image edge-blending and geometric calibration, and the key technologies such as tractors dynamics modeling, experiment reappearance and real-time test of virtual experiment were discussed. Finally, based on three-dimension geometry model and Vega Prime, the digitalization of tractor testing ground and real-time performance simulation were realized, which provides a new research means and technical method for the tractor performance testing. The test results show that the platform could make the observers immersed among the virtual environment to experience testing process directly and intuitively, and could achieve the real-time interaction between users and environment. This study proves stable operation of the system with reliability and validity, its velocity can reach 30 frames/s and the delay time is 0.025s.

Virtual Prototyping Analysis on Gear Assembly of the Power Train for a Large-scale Combine Harvester

Stabilization of the operating speed is a vital factor that affects a combine harvesters productivity. In this paper, the pretreatment of CAD models is set for precise simulation. Using CAE tools, the research concentrates on the static and kinematic characters of the gear assembly in the power train. Ultimately, the comparisons of the simulation results with the theoretical ones are conducted. The results indicate that the designed gear assembly can remain the operating speed stable and reliable under the assumed condition.

Driver Behavior Failure Probability Prediction Based on CREAM

As the major factor to driving safety, human error is induced both by inner factors and external environment factors. But until now, the systematical method for predicting driver’s behavior failure has not been put forward yet. Therefore, based on the contextual control model used in CREAM, the effects of the driving environment to the driver’s behavior are emphasized, and the characters in driving process, as well as the common performance conditions in driving environment are analyzed. The modified coefficient of the common performance conditions on the driver’s behavior failure is put forward. At the specific driving situation, the probable failure model of the driving operation is analyzed, and the failure probability of the driver’s behavior is predicted. The example shows that this method can be used simply, the prediction result can reflect the fact basically, and it can be used as reference for further research on driver reliability. Keywords-CREAM; human error; driver behavior; failure probability prediction