检索结果-内蒙古大学图书馆

Research on Person Recognition Model of Domain Adaptive Learning

学校读者我要写书评

暂无评论

Research on Person Recognition Model of Domain Adaptive Lear...

第32届中国控制与决策会议

作者： Ming-hai Jiao Wei-ming Duan Jue Wang Ben-dong Luo Chi Zhang Xiang-yu Sun School of Computer Science and Engineering Northeastern University Faculty of Robot Science and Engineering Northeastern University College of Science Northeastern University School of Mechanical Engineering and Automation Northeastern University

The person recognition problem in different domain is a hard challenge on lower accuracy of object detection with massive image data, that is the gap between person target and background is small and hard to distinguish different posture of person target. The novel domain adaptive person detection model based on transfer data training is proposed for detecting data on new target domain. And domain adaptive loss function is designed to control the difference between the two models with self training, so as to avoid catastrophic forgetting in the source domain. And then the experiment results show that the improved deep self training domain adaptive learning model and loss function algorithm are effective and efficient for person recognition problem.

关键词： Person recognition Deep self training Domain adaptive learning Loss function

Diagnostic Posture Control System for Seated-Style Echocardiography robot

学校读者我要写书评

暂无评论

arXiv 2022年

作者： Shida, Yuuki Sugawara, Masami Tsumura, Ryosuke Chiba, Haruaki Uejima, Tokuhisa Iwata, Hiroyasu Graduate School of Creative Science and Engineering Waseda University Tokyo169-8050 Japan Global Robot Academia Laboratory Waseda University Tokyo169-8050 Japan NSK Ltd Technology Development Department New Field Products Development Center Technology Development Division Headquarters Kanagawa251-8501 Japan The Cardiovascular Institute Tokyo106-0031 Japan Faculty of Science and Engineering Waseda University Tokyo169-8050 Japan

Purpose Conventional robotic ultrasound systems were utilized with patients in supine positions. Meanwhile, the limitation of the systems is that it is difficult to evacuate the patients in case of emergency (e.g., patient discomfort and system failure) because the patients are restricted between the robot system and bed. Then, it is ideal that the patient undergoes the examination in the sitting position in terms of safety. Therefore, we validated a feasibility study of seated-style echocardiography using a robot. Method Preliminary experiments were conducted to verify the following two points: (1) the possibility of obtaining cardiac disease features in the sitting posture as well as in the conventional examination, and (2) the relationship between posture angle and physical burden. For reducing the physical burden, two unique mechanisms were incorporated into the system: (1) a leg pendulum base mechanism to reduce the load on the legs when the lateral bending angle increases, and (2) a roll angle division by a lumbar lateral bending and thoracic rotation mechanisms. Results Preliminary results demonstrated that adjusting the diagnostic posture angle enabled us to obtain the views, including cardiac disease features, as in the conventional examination. The results showed that the body burden increased as the posture's lateral bending angle increased. The results also demonstrated that the body load reduction mechanism incorporated in the results could reduce the physical load in the seated echocardiography. Conclusion These results showed the potential of the seated-style echocardiography robot. © 2022, CC BY.

关键词： Human robot interaction

Shape-Adaptive Planning and Control for a Deformable Quadrotor

学校读者我要写书评

暂无评论

arXiv 2025年

作者： Wu, Yuze Han, Zhichao Wu, Xuankang Zhou, Yuan Wang, Junjie Fang, Zheng Gao, Fei State Key Laboratory of Industrial Control Technology Zhejiang University Hangzhou310027 China Huzhou Institute Zhejiang University Huzhou313000 China Faculty of Robot Science and Engineering Northeastern University Shenyang110819 China

Drones have become essential in various applications, but conventional quadrotors face limitations in confined spaces and complex tasks. Deformable drones, which can adapt their shape in real-time, offer a promising solution to overcome these challenges, while also enhancing maneuverability and enabling novel tasks like object grasping. This paper presents a novel approach to autonomous motion planning and control for deformable quadrotors. We introduce a shape-adaptive trajectory planner that incorporates deformation dynamics into path generation, using a scalable kinodynamic A* search to handle deformation parameters in complex environments. The backend spatio-temporal optimization is capable of generating optimally smooth trajectories that incorporate shape deformation. Additionally, we propose an enhanced control strategy that compensates for external forces and torque disturbances, achieving a 37.3% reduction in trajectory tracking error compared to our previous work. Our approach is validated through simulations and real-world experiments, demonstrating its effectiveness in narrow-gap traversal and multi-modal deformable tasks. Copyright © 2025, The Authors. All rights reserved.

关键词： Shape optimization

Manipulability of the Kondo effect in a T-shaped triple-quantum-dot structure

学校读者我要写书评

暂无评论

Physical Review B 2020年第8期102卷 085418-085418页

作者： Guang-Yu Yi Cui Jiang Lian-Lian Zhang Su-Rui Zhong Hao Chu Wei-Jiang Gong College of Sciences Northeastern University Shenyang 110819 China Basic Department Shenyang Institute of Engineering Shenyang 110136 China Faculty of Robot Science and Engineering Northeastern University Shenyang 110819 China

We theoretically investigate the Kondo effect of a T-shaped triple-quantum-dot structure, by means of the numerical renormalization group method. It is found that the system's entropy and conductance have opportunities to exhibit abundant transition processes for different interdot-coupling manners, with the decrease of temperature. All these results are caused by different Kondo physics mechanisms. Moreover, in the presence of appropriate parameters, the three-stage Kondo effect comes into being. Next when the electron-hole symmetry is broken or the structural parameters are changed, the Kondo resonance can also be observed in the conductance spectrum. However, it shows alternative dependence on the relevant quantities, i.e., the Coulomb interaction and interdot couplings. All these phenomena exhibit interesting Kondo physics in this system. We believe that this work can be helpful for further understanding of the Kondo effect in the triple-quantum-dot structures.

关键词： Kondo effect Quantum transport Renormalization group

Semantic dense reconstruction with consistent scene segments

学校读者我要写书评

暂无评论

arXiv 2021年

作者： Wan, Yingcai Li, Yanyan You, Yingxuan Guo, Cheng Fang, Lijin Tombari, Federico Dept. Faculty of Robot Science and Engineering Northeastern University Shenyan China Dept. Computer Science Technical University of Munich Munich Germany Key Laboratory of Machine Perception Peking University Beijing China MetaSpatie Tech Google Inc.

In this paper, a method for dense semantic 3D scene reconstruction from an RGB-D sequence is proposed to solve high-level scene understanding tasks. First, each RGB-D pair is consistently segmented into 2D semantic maps based on a camera tracking backbone that propagates objects' labels with high probabilities from full scans to corresponding ones of partial views. Then a dense 3D mesh model of an unknown environment is incrementally generated from the input RGB-D sequence. Benefiting from 2D consistent semantic segments and the 3D model, a novel semantic projection block (SP-Block) is proposed to extract deep feature volumes from 2D segments of different views. Moreover, the semantic volumes are fused into deep volumes from a point cloud encoder to make the final semantic segmentation. Extensive experimental evaluations on public datasets show that our system achieves accurate 3D dense reconstruction and state-of-the-art semantic prediction performances simultaneously. © 2021, CC BY.

关键词： Semantics

TP-TIO: A robust thermal-inertial odometry with deep thermalpoint

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Zhao, Shibo Wang, Peng Zhang, Hengrui Fang, Zheng Scherer, Sebastian Robotics Institute Carnegie Mellon University United States Faculty of Robot Science and Engineering Northeastern University China

关键词： Smoke

Lower limbs exoskeleton control strategy based on optimized compliance algorithm

学校读者我要写书评

暂无评论

Lower limbs exoskeleton control strategy based on optimized ...

2019 IEEE International Conference on robotics and Biomimetics, ROBIO 2019

作者： Wang, Fei Yu, Lignzhou Wu, Han Shi, Ketao Northeastern University Faculty of Robot Science and Engineering Shenyang Liaoning Province China Northeastern University College of Information Science and Engineering Shenyang Liaoning Province China

ISBN: (纸本)9781728163215

Aiming at achieving the compliance control of lower limbs exoskeleton, a compliant control strategy basing on optimized compliance algorithm was proposed. The motion trajectory of human body was collected by using inertial sensors and electromyography sensors, and the information is taken as the reference information to the exoskeleton robot. By introducing muscle activation degree as the feedforward loop to offset the hysteresis phenomenon which is shown in the process of movement, the following movement achieved compliant. The feasibility of this control method is both verified on the simulation software platform and physical prototype. The experimental results show that this control method is effective and feasible in the lower limbs exoskeleton system. © 2019 IEEE.

关键词： Compliance control

Event-VPR: End-to-end weakly supervised network architecture for event-based visual place recognition

学校读者我要写书评

暂无评论

arXiv 2020年

作者： Kong, Delei Fang, Zheng Li, Haojia Hou, Kuanxu Coleman, Sonya Kerr, Dermot College of Information Science and Engineering Northeastern University Shenyang China Faculty of Robot Science and Engineering Northeastern University Shenyang China Faculty of Computing Engineering and Built Environment Ulster University Northern Ireland United Kingdom

Traditional visual place recognition (VPR) methods generally use frame-based cameras, which is easy to fail due to dramatic illumination changes or fast motions. In this paper, we propose an end-to-end visual place recognition network for event cameras, which can achieve good place recognition performance in challenging environments. The key idea of the proposed algorithm is firstly to characterize the event streams with the EST voxel grid, then extract features using a convolution network, and finally aggregate features using an improved VLAD network to realize end-to-end visual place recognition using event streams. To verify the effectiveness of the proposed algorithm, we compare the proposed method with classical VPR methods on the event-based driving datasets (MVSEC, DDD17) and the synthetic datasets (Oxford robotCar). Experimental results show that the proposed method can achieve much better performance in challenging scenarios. To our knowledge, this is the first end-to-end event-based VPR method. The accompanying source code is available at https://***/kongdelei/Event-VPR. Copyright © 2020, The Authors. All rights reserved.

关键词： Cameras

Funnel nonsingular terminal sliding mode control based on TDE for robotic manipulators with model uncertainties

学校读者我要写书评

暂无评论

Funnel nonsingular terminal sliding mode control based on TD...

Chinese Automation Congress (CAC)

作者： Huaizhen Wang Lijin Fang Tangzhong Song Menghui Hu Zhenghan Wu Faculty of Robot Science and engineering Northeastern University Shenyang China Cheeloo college of medicine Shandong University Jinan China

ISBN: (数字)9781728176871

ISBN: (纸本)9781728176888

Due to the accurate robotics dynamics model or numerous parameters are usually unsuitable for real applications, a funnel nonsingular terminal sliding mode control(NTSMC) based on time delay estimate(TDE) is developed in our research paper. Firstly, the TDE technique is developed to compensate the system dynamics for robotic manipulators, which can achieve an ideal model-free advantage. Further, a funnel variable is used to transform the tracking error fall within funnel boundary. Based on the funnel variable, a novel funnel NTSMC is designed to provide transient and steady-state trajectory tracking performance for robot system. Finally, the reasonable experimental results on Rethink sawyer are presented to verify the high accurate of trajectory tracking performance of our researched approach.

关键词： robots Manipulator dynamics Service robots Trajectory tracking Delay effects Estimation Transient analysis