Fast object detection based on binary deep convolution neural networks

In this study,a fast object detection algorithm based on binary deep convolution neural networks(CNNs)is proposed.Convolution kernels of different sizes are used to predict classes and bounding boxes of multi-scale objects directly in the last feature map of a deep CNN.In this way,rapid object detection with acceptable precision loss is achieved.In addition,binary quantisation for weight values and input data of each layer is used to squeeze the networks for faster object detection.Compared to full-precision convolution,the proposed binary deep CNNs for object detection results in 62 times faster convolutional operations and 32 times memory saving in theory,what’s more,the proposed method is easy to be implemented in embedded computing systems because of the binary operation for convolution and low memory requirement.Experimental results on Pascal VOC2007 validate the effectiveness of the authors’proposed method.

Application of BIM Technology in Rapid Modeling of Residential Building Design

Combining with engineering examples,this paper introduces BIM into architectural design model and adopts Revit rapid modeling.It focuses on realizing rapid combination modeling of similar components through secondary development,improving information conversion efficiency between engineering CAD and Revit building structure modeling,and quickly generating BIM structure construction drawings.

Alström syndrome with a novel mutation of ALMS1 and Graves’hyperthyroidism:A case report and review of the literature

BACKGROUND Alström syndrome(AS,OMIM ID 203800)is a rare disease involving multiple organs in children and is mostly reported in non-Chinese patients.In the Chinese population,there are few reports on the clinical manifestations and pathogenesis of AS.This is the first report on the association between AS and Graves’hyperthyroidism.CASE SUMMARY An 8-year-old Chinese girl was diagnosed with AS.Two years later,Graves’hyperthyroidism developed with progressive liver dysfunction.The patient’s clinical data were collected;DNA from peripheral blood of the proband,parents and sibling was collected for gene mutation detection using the second-generation sequencing method and gene panel for diabetes.The association between the patient’s genotype and clinical phenotype was analyzed.She carried the pathogenic compound heterozygous mutation of ALMS1(c.2296_2299del4 and c.11460C>A).These stop-gain mutations likely caused truncation of the ALMS1 protein.CONCLUSION The manifestation of hyperthyroidism may suggest rapid progression of AS.

An Overview of Contour Detection Approaches

Object contour plays an important role in fields such as semantic segmentation and image classification. However, the extraction of contour is a difficult task, especially when the contour is incomplete or unclosed. In this paper, the existing contour detection approaches are reviewed and roughly divided into three categories： pixel-based, edge-based, and region-based. In addition, since the traditional contour detection approaches have achieved a high degree of sophistication, the deep convolutional neural networks （DCNNs） have good performance in image recognition, therefore, the DCNNs based contour detection approaches are also covered in this paper. Moreover, the future development of contour detection is analyzed and predicted.

Skill Learning for Robotic Insertion Based on One-shot Demonstration and Reinforcement Learning

In this paper,an efficient skill learning framework is proposed for robotic insertion,based on one-shot demonstration and reinforcement learning.First,the robot action is composed of two parts:expert action and refinement action.A force Jacobian matrix is calibrated with only one demonstration,based on which stable and safe expert action can be generated.The deep deterministic policy gradients(DDPG)method is employed to learn the refinement action,which aims to improve the assembly efficiency.Second,an episode-step exploration strategy is developed,which uses the expert action as a benchmark and adjusts the exploration intensity dynamically.A safety-efficiency reward function is designed for the compliant insertion.Third,to improve the adaptability with different components,a skill saving and selection mechanism is proposed.Several typical components are used to train the skill models.And the trained models and force Jacobian matrices are saved in a skill pool.Given a new component,the most appropriate model is selected from the skill pool according to the force Jacobian matrix and directly used to accomplish insertion tasks.Fourth,a simulation environment is established under the guidance of the force Jacobian matrix,which avoids tedious training process on real robotic systems.Simulation and experiments are conducted to validate the effectiveness of the proposed methods.

Motion Control in Saccade and Smooth Pursuit for Bionic Eye Based on Three-dimensional Coordinates

Saccade and smooth pursuit are two important functions of human eye. In order to enable bionic eye to imitate the two functions, a control method that implements saccade and smooth pursuit based on the three-dimensional coordinates of target is proposed. An optimal observation position is defined for bionic eye based on three-dimensional coordinates. A kind of motion planning method with high accuracy is developed. The motion parameters of stepper motor consisting of angle acceleration and turning time are computed according to the position deviation, the target＇s angular velocity and the stepper motor＇s current angular velocity in motion planning. The motors are controlled with the motion parameters moving to given position with desired angular velocity in schedule time. The experimental results show that the bionic eye can move to optimal observation positions in 0.6 s from initial location and the accuracy of 3D coordinates is improved. In addition, the bionic eye can track a target within the error of less than 20 pixels based on three-dimensional coordinates. It is verified that saccade and smooth pursuit of bionic eye based on three-dimensional coordinates are feasible.

A New Diagnosis Method with Few-shot Learning Based on a Class-rebalance Strategy for Scarce Faults in Industrial Processes

For industrial processes, new scarce faults are usually judged by experts. The lack of instances for these faults causes a severe data imbalance problem for a diagnosis model and leads to low performance. In this article, a new diagnosis method with few-shot learning based on a class-rebalance strategy is proposed to handle the problem. The proposed method is designed to transform instances of the different faults into a feature embedding space. In this way, the fault features can be transformed into separate feature clusters. The fault representations are calculated as the centers of feature clusters. The representations of new faults can also be effectively calculated with few support instances. Therefore, fault diagnosis can be achieved by estimating feature similarity between instances and faults. A cluster loss function is designed to enhance the feature clustering performance. Also, a class-rebalance strategy with data augmentation is designed to imitate potential faults with different reasons and degrees of severity to improve the model’s generalizability. It improves the diagnosis performance of the proposed method. Simulations of fault diagnosis with the proposed method were performed on the Tennessee-Eastman benchmark. The proposed method achieved average diagnosis accuracies ranging from 81.8% to 94.7% for the eight selected faults for the simulation settings of support instances ranging from 3 to 50. The simulation results verify the effectiveness of the proposed method.

An Effective On-line Surface Particles Inspection Instrument for Large Aperture Optical Element

Surface particles growing in large aperture optical element （LAOE） have significant impact on LAOE＇s stable operation. It is a challenge for the online system to inspect the particles with long working distance, enough precision and high efficiency because of the system constraints. In this paper, an effective and portable inspection instrument is designed based on dark-field imaging principle. A Nikon lens and an industrial high definition （HD） camera are selected to construct the vision system to inspect particles of microns size spreading over hundreds of millimeters. Using two motors and other mechanical structure, the system can realize auto-focus and image rectification functions. The line light sources are installed on both sides of the LAOE in a sealed box while the vision system is portable and working outside the box. An adaptive binarization method is proposed to process the captured dark-field image. The distribution of particles on the LAOE＇s surface is investigated. Because of the high resolution of the captured image, the SSE2 instructions optimization method is used to reduce the time cost of the algorithm. Experiments show that the instrument can inspect LAOE effectively and accurately.

Viewpoint Planning for Freeform Surface Inspection Using Plane Structured Light Scanners

This paper proposes an automatic model-based viewpoint planning method, which can achieve high precision and high efficiency for freeform surfaces inspection using plane structured light scanners. The surface model is utilized in stereolithography format, which is widely used as an industrial standard. The proposed method consists of 4 steps： topology reconstruction, mesh refinement, scan direction determination and viewpoint generation. In the first step, the topology structure of the surface model is reconstructed according to a designed data structure, based on which a neighborhood search algorithm is developed. In the second step, big facets in the surface model are segmented into several small ones, which are suitable for viewpoint planning. In the third step, an initial scan region of a viewpoint is grouped by the neighborhood search algorithm combining with total area and normal vector restrictions. Accordingly, the scan direction is determined by the normal vectors of facets in the initial scan region. In the fourth step, the position, the orientation, and the final scan region of the viewpoint are determined by 4 scan constraints, i.e., field of view, working distance range, view angle and overlap. Experimental results verify the effectiveness and advantages of the proposed method.

Position/Force Hybrid Control System for High Precision Aligning of Small Gripper to Ring Object

A position/force hybrid control system based on impedance control scheme is designed to align a small gripper to a special ring object. The vision information provided by microscope vision system is used as the feedback to indicate the position relationship between the gripper and the ring object. Multiple image features of the gripper and the ring object are extracted to estimate the relative positions between them. The end-effector of the gripper is tracked using the extracted features to keep the gripper moving in the field of view. The force information from the force sensor serves as the feedback to ensure that the contact force between the gripper and the ring object is limited in a small safe range. Experimental results verify the effectiveness of the proposed control strategy.

Orientation Measurement for Objects with Planar Surface Based on Monocular Microscopic Vision

Orientation measurement of objects is vital in micro assembly.In this paper,we present a novel method based on monocular microscopic vision for 3-D orientation measurement of objects with planar surfaces.The proposed methods aim to measure the orientation of the object,which does not require calibrating the intrinsic parameters of microscopic camera.In our methods,the orientation of the object is firstly measured with analytical computation based on feature points.The results of the analytical computation are coarse because the information about feature points is not fully used.In order to improve the precision,the orientation measurement is converted into an optimization process base on the relationship between deviations in image space and in Cartesian space under microscopic vision.The results of the analytical computation are used as the initial values of the optimization process.The optimized variables are the three rotational angles of the object and the pixel equivalent coefficient.The objective of the optimization process is to minimize the coordinates differences of the feature points on the object.The precision of the orientation measurement is boosted effectively.Experimental and comparative results validate the effectiveness of the proposed methods.

Realization of an Automated Microassembly Task Involving Micro Adhesive Bonding

An automated approach is proposed for a microassembly task, which is to insert a 10 μm diameter glass tube into a 12 μm diameter hole in a silicon substrate, and bond them together with ultraviolet （UV） curable adhesive. Two three-degree-of-freedom micromanipulators axe used to move the glass tube and the dispensing needle, respectively. Visual feedback is provided by an optical microscope. The angle of the microscope axis is precisely calibrated using an autofocus strategy. Robust image segmentation method and feature extraction algorithm are developed to obtain the features of the hole, the glass tube and the dispensing needle. Visual servo control is employed to achieve accurate aligning for the tube and the hole. Automated adhesive dispensing is used to bond the glass tube and the silicon substrate together after the insertion. On-line monitoring ensures that the diameter of the adhesive spot is within a desired range. Experimental results demonstrate the effectiveness of the proposed strategy.

Tracking and Guiding Multiple Laser Beams for Beam and Target Alignment

An accurate and robust approach for tracking and guiding multiple laser beams is developed, which can be applied to the task of beam and target alignment. Multiple laser spots are firstly detected and recognized from the image sequences of the target and laser spots. Then, the contour tracking algorithm based on the chain code is investigated, in which the shape matching scheme based on the invariant moments is employed to distinguish different spots. When occlusion occurs in the multiple spots tracking procedure,the contour tracking combined with Kalman filter prediction is proposed to obtain the positions of multiple spots in real-time. In order to guide 3 spots to align the target, an incremental proportional integral(PI) controller is employed to make the image features of spots converge to the desired ones. Comparative experiments show that, the proposed tracking method can successfully cope with the fast motion, partial or complete occlusion. The experiment results on spots guiding also exhibit the accurate and robust performance of the strategy. The proposed visual system solves the problem of spots mixing, reduces the alignment time, improves the shooting accuracy and has been successfully applied to the experimental platform.