Abstract:

Abstract:Robots are important equipment in modern industrial manufacturing and production. As products move towards small batch, multi-variety, and flexibility, robot collaboration systems based on multi-information fusion enable high-end precision manufacturing industries. This research focuses on the precision alignment and insertion technology of a hand-eye cooperation system in the field of precision electronic component assembly. By establishing a contact state model between electronic components and heterogeneous plates, the motion characteristics of force and displacement are analyzed. A compound control algorithm integrating visual, force sense, and encoder information is proposed, combined with visual inspection and tracking technology. Component insertion comparison experiments and assembly experiments based on algorithm fusion are carried out on the electronic component assembly platform. The results show that the positioning accuracy in the alignment phase is within 0.185 pixels. The contact state judgment and adjustment algorithm in the assembly stage ensures the safe and effective assembly of components.

Abstract:The existing helmet detection system mainly uses a fixed camera, it cannot achieve full-area detection, and the previous detection algorithms based on deep learning have complex structures and high computational costs, which cannot meet the requirements of using mobile vehicles and embedded devices. In this paper, a lightweight helmet detection algorithm scheme based on unmanned aerial vehicle is proposed. The drone is loaded with camera to collect images of the construction site, and the image data is transferred to the computer via wireless communication. Based on the YOLOv5s target detection algorithm, a lightweight detection algorithm is investigated. To improve the detection more efficient, the YOLOv5s target detection algorithm is improved in terms of multi-scale detection, image preprocessing, unbalanced positive and negative samples, and inference speed. This design scheme combines deep learning and unmanned aerial vehicle technology, not only to realize real-time automatic detection of helmet wearing, but also can realize the full-area helmet detection of the construction site. Real experiments show that, the lightweight target detection model is only 1.72% lower than the mean average precision of the original model. The inference speed on the same CPU can be doubled, and the floating-point calculation is reduced from 16.5 billion to 3.4 billion times per second. The model size is almost 1/10 of the original size.

Abstract:A multi-objective rapidly-exploring random tree path optimization method is proposed for the multiobjective patrol path planning of mobile robots. According to the provided environment map and patrol target points, a new method RRT-Connect-ACO is used to obtain the patrol sequence and feasible path of the target points. Then the optimal path is obtained by introducing informed subset to optimize the path. The experiment results show that the method considers the influence of terrain and obtains an optimal path that is more consistent with the actual situation, which is different from the existing multi-objective path planning algorithms.

Abstract:With the rapid development of related technology, the virtual reality is attracting increasing attention. As an intuitive object for human-computer interaction, the virtual human plays an important role in the virtual environment. During the process of building plausible virtual human, one crucial step is to create emotional facial expressions. The mainstream methods often rely on hand-crafted efforts by designers, resulting a laborious and time-consuming task. To address this problem, this paper introduces a method that generates emotional expressions based on manipulating blendshapes. Given an arbitrary facial expression image, this method estimates the corresponding blendshape coefficients, which can be used to generate the target emotional expression on the virtual human face. Experiment results show that the proposed method has strong generalization ability and is effective on reducing the burden of human designers in the task of emotional expressions generation.

Abstract:Affected by crosswind, vehicles under high speed are liable to deviate from the expected trajectory, which would cause higher risk of misoperation of the driver and considerable safety hazard. For the consideration of the above situation, the investigation of active control of the vehicle crosswind stability was conducted. In this research, a 3 degrees of freedom vehicle dynamic with aerodynamic force simulation model was established, a prediction controller of the vehicle crosswind stability of an active front-wheel steering vehicle was designed, and a Simulink-CarSim co-simulation platform was built for validation analysis. Results reveal that, under the working condition of unidirectional and multi-directional, the vehicle with crosswind stability controller possesses the biggest deviation value of 0.01 m, which is extremely lower than the value obtained from the vehicle without crosswind stability controller while the platform value of yaw velocity is kept around zero, and the peak value of which is reduced by 80%, which means the crosswind stability was remarkably raised.

Abstract:Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by impaired social communication, repetitive behaviors, and restricted interests. Signs of autism usually appear by age 3, but the cinical diagnosis is time-consuming and subjective. Therefore, a rapid and cost-effective assessment method is urgently needed. Children with ASD have atypical gaze patterns when they precept emotional stimuli, which suggests a great potential use of eye-tracking technology as an assessment method for ASD detection. This paper proposes a model for automatically assessing children with ASD based on atypical gaze patterns. The model extracts the eye movement trajectory features of perceived emotions in natural scenes, and uses the machine learning model to learn to automatically identify ASD according to the eye movement trajectory features. Results show that the accuracy reaches 79.71%, which has potentially become an early ASD children screening approach.

Abstract:Terahertz wave is the electromagnetic field with frequencies between that of infrared and millimeter waves. It has attracted increasing attention in both fundamental research and technological applications for its unique characteristics. In terahertz telecommunication, radar and imaging systems, terahertz antenna is key for the performance. To date, reported terahertz antennas suffer from limited phase modulation, relatively low efficiency and small beaming angle. Here these challenges are addressed by first designing three types of terahertz graphene antennas with the smallest size of 5 μm. Then a novel type of antenna with dual resonances is proposed, and achieve dynamic phase modulation within the full 2π range and meanwhile high efficiency above 20%. This performance surpasses antennas with single resonance since dual resonances alleviate the contradiction between large phase modulation range and high efficiency. Terahertz graphene antennas are fabricated with standard micro-fabrication processes, and experimentally obtained terahertz dynamic phase modulation of 1.03 THz with reflection efficiency above 23%, which agree basically with simulation results. Making use of the phase modulation metasurfaces tuned according to continuous phase coding, terahertz beam steering is numerically realized with dynamic range of －25°～25°. This work will provide a strategy for achieving large-range phase modulation and beam steering beyond the terahertz regime.

Abstract:In augmented reality, virtual reality and the metaverse, three dimensional (3D) reconstruction technologies play important roles in acquiring the content information. Among them, the structured light method has been widely used due to the advantages of high precision and not being affected by the texture of the surface material on the object. Traditional structured light 3D reconstruction mainly uses digital light processing based projectors to project coded patterns. However, the shortcomings of digital light processing projectors such as large size, high power, and high price limit their applications. Therefore, more 3D scanning systems start to use a micro-electro-mechanical system mirror as a structured light projector, which has small size, low cost, and high frame rate. In this paper, the phase-height model is proposed to complete the 3D scanning system based on micro-electro-mechanical system mirror. To deal with the noise caused by the speckle effect, anti-noise performance of three time-phase unwrapping algorithms commonly used for micro-electro-mechanical system mirror are experimentally compared. Results show that the multi-frequency hierarchical method and the negative exponential fitting method show better anti-noise capability and higher precision, while the multi-frequency heterodyne method shows poor anti-noise performance. This study provides a guide in choosing proper phase resolution method for micro-electro-mechanical system based 3D scanning systems.