Search

Robot soccer games are important activities for the research of the system design and intelligent control, but the actual implementation process is often limited to the actual quality and quantity of equipment. A testing method based on virtual prototype software is proposed for the single robot in soccer games on the basis of theoretical analysis of the robot soccer system. A real-time interaction method is realized between the control algorithm and system model. According to the position of the ball and robot fed back from the model, the control program can plan appropriate trajectory and sends command to the left and right wheel, which realizes the real-time tracking of the ball. The simulation and actual combating results show that the method can be used for the simulation of the actual robot soccer games and the optimization of the control algorithm.

Underwater pipelines of oil and gas need periodic inspection to prevent damage due to the biological activity of water, turbulent current and tidal abrasion. Currently, vision-based autonomous underwater vehicle plays an important role in this field. A system has been designed to help an autonomous vehicle in sea-bottom survey operation. Image understanding and object recognition directly affect the accuracy of inspection. An image smoothing method based on mathematical morphology is proposed. The disturbances on acquired images caused by the motion are partially removed. A series of algorithms about image preprocessing, segmentation and recognition are proposed to access pipeline contours from the top-view images effectively. Navigation data based on Hough transformation is presented after the analysis of contours. Finally, the processing effect on a pipeline image demonstrates the effectiveness of the system.

A template matching method based on a vision system is proposed to get the location of terminal blocks. Gaussian pyramid decomposition is used to get the source image and template image samples, which will reduce the matching time and thus fulfill the requirement of real time application. The positioning screws used as mark points can be obtained by binarization, erosion, and dilation, and then the location of them can be worked out by the centre-of-gravity method. Finally, the location and the rotation angle of terminal blocks can be obtained with least square method. Experimental results show that this method is convenient to be operated with high recognition rate and efficiency.

The interaction force and the environments uncertainties are the most challenges for robotic material removal process. The conventional constant force control methods for the deburring process have the inherent characteristic of leaving the deburred surface as an imprint of the original. A process force model considering the burrs variation is presented to predict the contact force in robotic machining process. A self-tuning fuzzy strategy is adopted to implement the on-line compensation for the static error caused by the traditional impedance controller. The fuzzy controller is adjusted by an updating factor to select the most appropriate fuzzy rule set based on the measured performance results. Simulation results show efficacy of the proposed method in robotic machining process, and the control performance is better than that of a traditional impedance controller.