Abstract:Deep space exploration aims at the exploration of planets,moons,asteroids,comets and space environment in the universe other than the Earth.It is an important task to properly select scientific instruments for deep space exploration,which enable human to figure out the physical and chemical environment,topography,dynamic characteristics and the existence of organic matter and water.It is of great significance for mankind to explore the existence of extraterrestrial water and life,as well as the origin and evolution of the solar system.This paper summarizes the scientific objectives of the deep space exploration,then introduces the scientific instruments and their characteristics for typical deep space exploration missions in recent 50 years,including mass spectrometer,spectrometer,ground penetrating radar and magnetometer,and sorts out their performance parameters,the exploration oriented missions and tasks of the instruments.Finally,this paper discusses the development trend of scientific instruments for deep space exploration,such as miniaturization,low power consumption,stronger environmental adaptability and the combination of multiple instruments.

Abstract:Visual Odometry (VO),which is an important part of visual simultaneous localization and mapping technology,is mainly used for robot pose estimation and local map building through time series images captured by camera sensors.Known as the front end,VO has been widely used in many fields and achieved fruitful practical results,and it is of great significance for unmanned driving,autonomous drones,virtual reality,and augmented reality,etc.In this paper,we summarize the recent research results on the novel visual odometry technology based on introduction of various algorithms in the framework module of classical VO.According to their technical means,the novel methods are divided into two categories,including VO integrated with multiple sensors (take VIO as an example),and VO based on deep learning.The former improves the accuracy of VO by complementing the advantages of various sensors,while the latter is combined with deep learning network.Finally,the existing algorithms of visual odometry are compared,and the future development trend is forecasted based on the challenges faced by VO.

Abstract:Autonomous operation of vision-based manipulator is of great significance in many fields such as industrial production,aerospace and deep-sea exploration.Here we summarize the autonomous operation technology of vision-based manipulator in light of their control schemes:geometric driven control and data driven control.The control principles as well as related work are introduced,then strengths of and challenges faced by each category are analyzed.Finally,key issues and challenges for autonomous control of vision-based manipulator in fields of ocean development and exploration are summarized.

Abstract:As an effective means of human-computer interaction, gesture recognition has become a hot topic in current research. In order to solve the problems of spatio-temporal variability and feature complexity concerning dynamic gestures, we propose a dynamic gesture recognition solution based on 3D skeleton features. The accuracy of dynamic gesture recognition is greatly impaired due to the temporal differences and complexity of dynamic gestures, thus a key frame extraction algorithm is designed to extract key features of dynamic gestures for further feature extraction. To overcome the difference in classification performance between single classifiers, multiple classifiers are used to simultaneously classify and fully exploit gesture features. We also propose an adaptive fusion algorithm to effectively fuse different classifiers according to their classification performances thus improve the final classification accuracy. Finally, experiments are carried out, and results verify the effectiveness of the proposed dynamic gesture recognition approach.

Abstract:Ant colony algorithm has slow convergence rate,low efficiency and often gets local optimal solution.We propose an adaptive way to change the amount of pheromones,which can speed up the convergence rate.We also improve the heuristic function to increase the purpose of ant colony search,as well as reduce the probability of falling into local optimal solution.Simulations are carried out to verify the effectiveness of the proposed algorithm,and the results show that the global optimal search ability and convergence rate are greatly improved.

Abstract:To make up for the shortage of experimental devices for the emerging majors such as intelligent manufacturing and robot engineering,as well as the urgent needs in improving students' innovative and practical ability,we design a dual-arm cooperative teaching robot based on binocular vision.The mechanical part of the two arms is a symmetrical structure mounted on a big stable base to reduce manufacturing cost and coordinated grasping difficulty.In order to increase the accuracy of the robot terminal pose acquisition and thus improve the dual-arm cooperative grasp in both accuracy and stability,we design a teaching motion restoring algorithm and a reprojection error minimization algorithm.The upper computer of the system can drag and control the two arms to do programming teaching and action group experiments;and the robot can not only process images and conduct single-arm grabbing experiments under visual guidance,but also carry out motion analysis experiments,online programming,and open experiments of coordinated grabbing.The proposed system is cost-effective,rich in functions,highly integrative,and able of both drag teaching and open experiments,which is helpful to improve students' innovative and practical ability,therefore it is very suitable for promotion and application in the experimental teaching courses for intelligent manufacturing,robot engineering and other related majors.

Abstract:A patrol robot system is designed and implemented in order to enrich the means of safety inspection and improve the ability of emergency response in hazardous chemical warehouses in university.To build the patrol robot platform,multiple sensors including odometry,single line scan laser rangefinder,night vision color and thermal imaging cameras with pan-tilt unit,and environmental sensor suite are integrated and boarded on a differentially driven four-wheel mobile chassis.Data collected from odometry and laser rangefinder are merged and employed to realize Simultaneous Localization and Mapping (SLAM),patrol path definition as well as path following.At any patrol position,the pre-defined objects are located by image correlation analysis and their temperatures are measured by thermal imaging.In the patrol process,the thermal imaging camera and the environmental sensors are used to measure environmental parameters in real-time thus generate heat trajectories.Intruders which appear in the thermal images are detected based on prior knowledge.All the related algorithms are implemented into the robot operating system,while the web system for remote human-robot interaction is developed using NodeJS.The experiments of autonomous patrol are performed in environments with and without illumination.Experimental results demonstrate that the proposed patrol robot system can meet the demand of safety inspection in hazardous chemical warehouse in university.

Abstract:This study aims to evaluate a WAM robot-assisted rehabilitation training technology targeted for improving arm function from aspects of robot kinematic parameters and traditional medical assessment.The experimental platform with WAM robot is built to assist the upper limb hemiplegia patients in marking some simple shapes or doing some interesting games with their upper limbs.Data of trajectory and torque of the WAM robot are collected,and the trajectory tracking error and torque change before and after rehabilitation training are analyzed to evaluate the upper limb function.At the same time,physicians use Brunnstrom levels to appraise the rehabilitation performance.The results show that one month rehabilitation training reduces the robot trajectory tracking error and the torque change,as well as improves the upper limb function indicated by the reduction in random limb movement and rise in Brunnstrom level.This study provides evidence on the effectiveness of the WAM robot-assisted rehabilitation training in improving upper limb's motor function,which is superior to conventional rehabilitation training.

Abstract:In order to improve the comprehensive practical ability of students majoring in robot engineering,a teaching robot experimental platform which integrates machinary,electricity,control and image processing is designed and developed in this paper.As a platform integrates robot mechanical structure design,kinematics analysis,motion control and image processing,it is able of joint assembly and demonstration experiments,which avails students to understand robot's mechanical structure and working principle,and more importantly,it is developed as an open control experimental platform that can complete complex experiments such as robot kinematics analysis,drive control,trajectory planning,image processing,and object recognition & grasp under vision guidance.Strong in integration and manipulability,the designed experimental platform can stimulate students' learning interest and improve their engineering practical and innovative ability effectively.

Abstract:In the field of human-computer interaction,the position information of human hands is often directly used in the interpretation of interactive instructions and the calculation of interactive results,therefore,high-precision real-time hand position detection is an important basis for non-contact and natural human-computer interaction.In order to solve the problem of fluctuation and error in the 3D coordinates acquired by Kinect 2.0,a hand position detection algorithm based on mean processing of correlation points is proposed in this paper.Based on the depth information,the hand image is segmented using the human wrist as the segmentation threshold,and the human hand position information is processed by spatial and temporal average of the related points to improve the accuracy of position detection.The experimental results show that the algorithm based on correlation points mean processing is effective and the detection error is less than 5 mm,which can meet the basic requirements of human-computer interaction.

Abstract:With the advances of China's space exploration,the research of related technologies including mobile manned lunar rover has been an important task for China's future manned lunar landing and exploration.This paper presents the design of a control system based on a force feedback handle for the driving of the mobile manned lunar rover according to the comfort and working space of the astronauts.Mainly composed of a 2-DOF rotating structure,a hand lever and a DC motor with optical encoder,the force feedback handle can give force and touch sensing to astronauts under the specifically designed environment,and assist astronauts to drive the rover safely and effectively.The hardware of the control system receives the command signal and force feedback information from the upper computer,then feeds back the position data of the handle to the upper computer.The software system builds the lunar rover driving scene based on unity3D,in which virtual sensors are used to make driving decisions and calculate the corresponding force feedback information.The simulation test results show that the force feedback handle can well assist astronauts to drive the mobile manned lunar rover,and effectively improve the driving efficiency and security of the mobile manned lunar rover.

Abstract:An adaptive recognition method based on Tesseract-OCR engine is proposed to solve the problem of extracting and real-time recognition of specific table items in any region of complex invoices.First,the invoice image is preprocessed by OpenCV for filtering,adaptive threshold,etc.,to get a binary image.Then,the open operation in morphology is used to extract the global line segments and position of the table.The coordinates of the intersection points of the table is combined with the custom template to realize the adaptive adaptation between the table header and the content.Then the jTessBoxEditor is used to train and optimize the content of the table items,and finally the character recognition based on Tesseract-OCR is realized.The experimental results show that this method has high accurate recognition rate,supports the adaptive recognition of ROI (Region of Interest),and is highly available.

Abstract:The distortion characteristics of magnetic force line for magnetic levitation ball winding are analyzed,based on which the motion model of magnetic levitation ball with axial and horizontal degrees of freedom is constructed.Then the matching and mismatching interference in the system is reconstructed into matching interference using model transformation,and a new system state space equation is established.To address the difficulty in obtaining the velocity and acceleration of the air gap in the suspension air gap,as well as the real-time observation of interference,a sliding mode observer for air gap velocity and acceleration with disturbance reconstruction is proposed.Based on this observer,a sliding mode tracking controller is designed.Finally,a sliding mode state observation and tracking cooperative control (SMSOTC) simulation platform with disturbance reconstruction is built to verify the optimization of the proposed control strategy in dynamic response speed,tracking error,and anti-interference performance compared with traditional PID control.

Abstract:The three-level inverter is suitable for high voltage and large capacity occasions for its advantages such as simple topology,less devices,and small voltage stress on each power transistor.Compared with the legacy two-level inverter,three-level inverter has more output levels,lower voltage and current harmonic content,and better sine waveform.A Model Predictive Current Control (MPCC) scheme is proposed for Interior Permanent Magnet Synchronous Motor (IPMSM) to improve its current dynamic response speed,and a second-order delay compensation strategy is presented to solve the time delay in control system.Validity of the proposed MPCC scheme is confirmed by simulation.The simulation results showed that the IPMSM-MPCC driven by three-level inverter has smaller harmonic content in output current waveform,better dynamic and steady-state performance,compared with that driven by two-level inverter.

Abstract:The Empirical Mode Decomposition (EMD) denoising and Multifractal Detrended Cross-Correlation Analysis (MF-DCCA) have been combined to address the shortcomings of traditional portfolio models,thus a multiple fractal portfolio model under EMD denoising and MF-DCCA is proposed in this paper.The new model,abbreviated as EMD-Mean-MF-DCCA,is applied to a highly speculative cryptocurrency portfolio,which is then verified by out-of-sample test and analysis with rolling window technique.The results show that whether the cryptocurrency price is on upward or downward trend,the proposed EMD-Mean-MF-DCCA is significantly optimized in terms of profitability and Sharpe ratio compared with traditional portfolio models and non-denoised fractal portfolio models.Moreover,the portfolio investment scheme under the new model has better risk resistance capability when the price of cryptocurrency falls sharply.

Abstract:The relationship between solar radiation indices and power generation efficiency were quantitatively studied in the fishery-solar hybrid photovoltaic power station in Baoying,Jiangsu.The results showed that solar radiation significantly impacted the power generation efficiency with visible seasonal fluctuations.Specifically,the daily peak of grid connected power and the power generation capacity were larger and more stable in summer and autumn,compared with those in winter and spring.Obvious logarithmic/linear correlation was observed between solar radiation indices and the daily peak grid connected power/daily power generation capacity,respectively.The daily peak of grid connected power and power generation capacity were correlated positively with average radiation,max radiation,and accumulated radiation in each month.The max radiation imposed the strongest influence on the peak value of grid-connected power,while the available sunshine hours showed the strongest influence on the monthly power generation capacity.The study indicated that average radiation,max radiation,and accumulated radiation should be taken as major indices in evaluating solar energy utilization and photovoltaic power station development.