title band














Ö÷Ò³ | Main Page




























About us | Research Focus | Publications | Products | Projects | Our Team
 


























  

Projects

  The Current Projects  |   The Completed Projects    














 













  

  The Current Projects£º






























1. The basic research on 3D vision based perception and guidance for industrial robots

2017.01-2020.12

Project Funding: Key Project of National Natural Science Foundation of China

Contact£ºProf. Dr. Yang Cong

Introduction: The research contents of the project: 1) the system of 3D perception based on Panoramic infrared structured light; 2) online learning models and algorithms for advanced industrial robots; 3) 3D object recognition and locating for autonomous operation of advanced industrial robots; 4) scene understanding and human-robot safty.
































2. Multimode perception system for Tri-Co robots on complicated illuminations and weather

2017.01-2020.12
Project Funding: National Natural Science Foundation of China

Contact£ºProf. Dr. Jiandong Tian

Introduction: The complex variation of illumination and meteorological condition causes many problems to the autonomous task and the active safety of Human-robot coexistence, which have a negative impact on the robustness and environment adaption of vision algorithms. Finding effective and robust methods to deal with this problem is one of the most important researches in computer vision and the related fields. Our solution benefits from both hardware and software, by utilizing a multi-modal information (image data, polarization, and radar) to solve this problem. This multi-modal information can guarantee the robustness and real-time performance of robot perceptual system. The survey on the current research trends further shows that this proposed multi-modal based study will be a prospective research topic. We have currently achieved preliminary results for the proposed research, thus provide a solid foundation for further work. This proposed method will be of great significance for the research of active environment perception of Human-robot coexistence.
































3. Visual simulation system of object, background and algorithms
 
2017.01-2018.12
Project Funding:Industrial Society

Contact£ºDr. Zhi Han,  Prof. Dr. Yandong Tang































4. Deep Learning and robot vision for intelligent service robot















2016-2017
Project Funding: Industrial Society

Contact£ºProf. Dr. Jiandong Tian  Prof. Dr. Yandong Tang
































5. Machine vision for underwater robot















2016-2018
Project Funding: Chinese Academy of Sciences

Contact£ºProf. Dr. Yandong Tang
































6. Online Learning for industrial cognitive network
 
2016.01-2020.12

Project Funding: Key Project of National Natural Science Foundation of China

Contact£ºProf. Dr. Yang Cong































7. Illumination modeling and its invariant algorithms for computer vision















2015.01-2018.12
Project Funding: National Natural Science Foundation of China

Contact£ºDr. Jiandong Tian
















We will develop the research on the illumination problems in image processing based on theories of atmospheric physics,physical optics, imaging mechanism and from the view of characteristic analysis of physical imaging. The aim of this project is to set up new models, to propose novel algorithms for effective processing illuminant problems in image processing. Our research content includes the calculable illumination model for computer vision, reflectance spectroscopy calculation of an image, intrinsic image, color constancy algorithm and illumination converting of an image.
































8. Partial Shape Recognition from Natural Scene Images
















2015.01-2017.12
Project Funding: National Natural Science Foundation of China

Contact£ºDr. Huijie Fan

We focus on the feature inconsistency and the local scale calculation problems in partial shape recognition algorithms to design robust partial shape descriptors.
































9. The basic research on an intelligent environment for minimally invasive spinal surgery
 
2014.01-2018.12

Project Funding: National Natural Science Foundation of China

Contact£ºProf. Dr. Yandong Tang
















The application of Minimally Invasive Spinal Surgery(MISS) is limited by some scientific and technology problems, such as the narrowed operational field of vision, the absence of continuous observation of instrument in spine£¬the lack of objective security for operation and the radiation damage to the surgeon. To solve these problems,the purpose of this project focuses on research works to realize instrument realtime positioning in human body, operation process visualization,robot force feedbacking, risk evaluationg and warning.
































10. Online Learning for Robot Scene Understanding
 
2014.01-2017.12

Project Funding: National Natural Science Foundation of China

Contact£ºProf. Dr. Yang Cong















 
  For scene classification, we intend to propose an effective online learning theory and establish a unified online learning framework. For modeling, depending on the low rank property of real data, we design a convex optimization model for online metric learning, which can decrease the complexity of the model and overcome over-fitting, and we pursuit the expectation of the model to guarantee the converge and use the stochastic optimization to achieve an efficient solution. For scene representation, we fuse the spatio-temporal cues and design an adaptive feature selection model based on group sparsty theory. Depending on this research, we will solve the overfitting and non-converge issues for online metric learning algorithm in theory, and achieve online learning for big data.

































11. The system of automatic early forest smoke/fire detection and alarming

Contact£º Prof. Dr. Yandong Tang
             
Mr. Qunhui Zheng, MSc., Research Fellow
 















Specification£º
 
1)Tower-based automatic & reliable early recognition of forest smoke and fire; 2)Automatic recognition of smoke clouds and fire by day and night; 3)The Farthest monitoring at a distance of up to 5  km.
















firedetection















































logo