This paper presents an automated defect management system based on machine learning and computervision that detects and quantifies different types of defects in porcelain products. The system is developed in collabor...
详细信息
ISBN:
(纸本)9781538626269
This paper presents an automated defect management system based on machine learning and computervision that detects and quantifies different types of defects in porcelain products. The system is developed in collaboration with an industrial porcelain producer and integrates robots, artificial vision and machine learning. At present, in most of the companies involved in the porcelain industry, defect detection is performed manually by employees. An intelligent system for product monitoring and defect detection is very much needed. Our proposed system is implemented through a convolutional neural network which analyzes images of the products and predicts if the product is defective or not. Experimental evaluation on an image data set acquired at the industrial partner shows promising results. The proposed architecture will finally have a positive economic impact for the company by optimizing the production flow and reducing the production costs.
The thesis is a further contribution to the area of robotic person following that has been an on-going research at the intelligentrobots and Systems Laboratory. The goal of this research was to enhance the vision bas...
The thesis is a further contribution to the area of robotic person following that has been an on-going research at the intelligentrobots and Systems Laboratory. The goal of this research was to enhance the vision based robotic person following with the accelerometer and gyroscope using a standard cell-phone device. The signals that are generated by the cell phone on the person while the person walks provide information about the number of steps and the direction of the motion of the person. These sensor values are sent through a personal wireless network to the robot computer. The received signals are process using various signal processing techniques including the Kalman filter and peak detection algorithms to extract the number of steps taken by the person and the direction of the motion of the person. The processed signals are then used by a fuzzy logic algorithm to determine the distance between the robot and the person, thus identifying the location of the person with respect to the robot. The robot then follows the person based on the estimated location of the person. The person is also identified by a vision system as complementary and secondary method for person detection. The vision-based system develops a number of morphological operations to detect the person and segment it from the scene and other objects in the image. The complete system has been implemented on a Segway robot platform, and a number of tests have been performed, these tests include walking of different persons and recording the distances travelled, the estimation of the person's location by the system and its comparison with the actual recorded measurements. The results show that in majority of cases the system is successful in person following in simple environments. The thesis concludes by outlining a number of improvements that can be made to the system for more robust operation.
This paper presents an automated defect management system based on machine learning and computervision that detects and quantifies different types of defects in porcelain products. The system is developed in collabor...
详细信息
ISBN:
(纸本)9781538626276
This paper presents an automated defect management system based on machine learning and computervision that detects and quantifies different types of defects in porcelain products. The system is developed in collaboration with an industrial porcelain producer and integrates robots, artificial vision and machine learning. At present, in most of the companies involved in the porcelain industry, defect detection is performed manually by employees. An intelligent system for product monitoring and defect detection is very much needed. Our proposed system is implemented through a convolutional neural network which analyzes images of the products and predicts if the product is defective or not. Experimental evaluation on an image data set acquired at the industrial partner shows promising results. The proposed architecture will finally have a positive economic impact for the company by optimizing the production flow and reducing the production costs.
In this paper, we design an autonomous flight controller for height regulation and bang-bang controller for directional control of a light-weight flapping-wing micro air vehicle (FWMAV) with limited payload. We also p...
详细信息
The intelligent Ground Vehicle Competition (IGVC) is one of four, unmanned systems, student competitions that1were founded by the Association for Unmanned Vehicle Systems International (AUVSI). The IGVC is a multidisc...
详细信息
The computational framework, based on the conformal camera, is developedfor processingvisual information during smooth pursuit movements of a robotic eye. During smooth pursuit, the image of the tracked object remains...
详细信息
In this paper, we design an autonomous flight controller for height regulation and bang-bang controller for directional control of a light-weight flapping-wing micro air vehicle (FWMAV) with limited payload. We also p...
详细信息
ISBN:
(纸本)9781509037636
In this paper, we design an autonomous flight controller for height regulation and bang-bang controller for directional control of a light-weight flapping-wing micro air vehicle (FWMAV) with limited payload. We also present autonomous vision-based target tracking for a FWMAV equipped with a low-cost and light-weight first person view (FPV) camera. We construct a ground station, integrated with control and visionalgorithms, which performs the image processing and the computation of control inputs based on the acquired state variables from motion capture system. In addition, we employ a vision algorithm for a low-quality camera to detect a static target with the discussions on the techniques to improve the reliability of visual detection. Experimental results show satisfactory flight performance, achieving the height regulation and directional control, and autonomous vision-based target tracking.
Spherical stereo vision is a kind of stereo vision system built by fish-eye lenses, which discussing the stereo algorithms conform to the spherical model. Epipolar geometry is the theory which describes the relationsh...
详细信息
ISBN:
(纸本)9781628414967
Spherical stereo vision is a kind of stereo vision system built by fish-eye lenses, which discussing the stereo algorithms conform to the spherical model. Epipolar geometry is the theory which describes the relationship of the two imaging plane in cameras for the stereo vision system based on perspective projection model. However, the epipolar in uncorrected fish-eye image will not be a line but an arc which intersects at the poles. It is polar curve. In this paper, the theory of nonlinear epipolar geometry will be explored and the method of nonlinear epipolar rectification will be proposed to eliminate the vertical parallax between two fish-eye images. Maximally Stable Extremal Region (MSER) utilizes grayscale as independent variables, and uses the local extremum of the area variation as the testing results. It is demonstrated in literatures that MSER is only depending on the gray variations of images, and not relating with local structural characteristics and resolution of image. Here, MSER will be combined with the nonlinear epipolar rectification method proposed in this paper. The intersection of the rectified epipolar and the corresponding MSER region is determined as the feature set of spherical stereo vision. Experiments show that this study achieved the expected results.
Obstacle detection and localization behaviours have shown to work robustly in 2D perceived environment. With the progress of proximity sensors, precisely the emergence of 3D visiontechniques, it would be interesting ...
详细信息
Obstacle detection and localization behaviours have shown to work robustly in 2D perceived environment. With the progress of proximity sensors, precisely the emergence of 3D visiontechniques, it would be interesting to examine these motion controllers whether for 3D perceived environment. 2D basic algorithms for obstacle avoidance or robot location, needs reformulation in order to process data from 3D perception devices. In this work, we introduce a 3D obstacle detection controller, combined with particle filter localization. The obstacle detection control described in this paper address 3D obstacles with different shapes and the particle filter uses 3D environment data to correct the wheelchair localization based on his kinematics model.
暂无评论