A5101909869- Reinforcement Learning in Robotics
- Robot Manipulation and Learning
- Remote Sensing and Land Use
- Civil and Geotechnical Engineering Research
- Advanced Image and Video Retrieval Techniques
- EEG and Brain-Computer Interfaces
- Advanced Algorithms and Applications
- Muscle activation and electromyography studies
- Robotic Path Planning Algorithms
- Diabetic Foot Ulcer Assessment and Management
- Retinal Imaging and Analysis
- Robotics and Sensor-Based Localization
- Simulation and Modeling Applications
- CCD and CMOS Imaging Sensors
- Social Robot Interaction and HRI
- Advanced Numerical Analysis Techniques
- Video Surveillance and Tracking Methods
- Material Dynamics and Properties
- Advanced Neural Network Applications
- Image Enhancement Techniques
- Computational Drug Discovery Methods
- Neural dynamics and brain function
- Soft Robotics and Applications
- Robotic Locomotion and Control
- Geomechanics and Mining Engineering
Beijing Institute of Technology
2021-2025
Southwest University
2023-2025
Ministry of Education of the People's Republic of China
2023-2024
Nanjing Tech University
2020-2024
Wuhan Polytechnic University
2012-2024
Emory University
2023-2024
The People's Hospital of Guangxi Zhuang Autonomous Region
2020-2023
Central China Normal University
2010-2023
Jiangnan University
2023
Beijing Advanced Sciences and Innovation Center
2023
There has been a substantial amount of research involving computer methods and technology for the detection recognition diabetic foot ulcers (DFUs), but there is lack systematic comparisons state-of-the-art deep learning object frameworks applied to this problem. DFUC2020 provided participants with comprehensive dataset consisting 2,000 images training testing. This paper summarizes results by comparing learning-based algorithms proposed winning teams: Faster R–CNN, three variants R–CNN an...
Since space intelligent robots are not restricted by physiological conditions, it is an attractive choice for the development of automation technology to use them exploration and utilization. It currently key direction major powers over world. This paper first investigates robotic manipulators humanoid robot systems station applications reviews theories methods achieve large-range stable motion dexterous manipulation. Then, on-orbit satellite maintenance reviewed, related technologies...
Protein kinase 2 (CK2) is a potential target, and the coumarins were identified as attractive CK2 inhibitors. In this study, two models (CoMFA CoMSIA) established, their reliabilities supported by statistical parameters. From CoMFA CoMSIA models, hydrophobic hydrogen bonds play very important roles in interactions between inhibitors CK2, which confirmed sufficiently molecular docking. Furthermore, binding mode of at active sites was also investigated docking study. The hydroxyl position R(5)...
Thyroid-associated ophthalmopathy (TAO) is one of the most common orbital diseases that seriously threatens visual function and significantly affects patients' appearances, rendering them unable to work. This study established an intelligent diagnostic system for TAO based on facial images.Patient images data were obtained from medical records patients with who visited Shanghai Changzheng Hospital 2013 2018. Eyelid retraction, ocular dyskinesia, conjunctival congestion, other signs noted...
Teleoperation enables robots to perform tasks in dangerous or hard-to-reach environments on behalf of humans, but most methods lack operator immersion and compliance during grasping. To significantly enhance the operator’s sense achieve more compliant adaptive grasping objects, we introduce a novel teleoperation method for dexterous robotic hands. This integrates finger-to-finger force vibrotactile feedback based Fuzzy Logic-Dynamic Compliant Primitives (FL-DCP) controller. It employs fuzzy...
Based on basic emotion modulation theory and the neural mechanisms of generating complex motor patterns, we introduce a novel emotion-modulated learning rule to train recurrent network, which enables musculoskeletal arm robotic perform goal-directed tasks with high accuracy efficiency. Specifically, inspired by fact that emotions can modulate process decision making through neuromodulatory system, present model generation adjust parameters adaptively, including reward prediction error, speed...
On-orbit assembly has become a crucial aspect of space operations, where the manipulator frequently and directly interacts with objects in complex process. The traditional control limitations adapting to diverse tasks is vulnerable vibration, leading failure. To address this issue, we propose human-like variable admittance method based on damping characteristics human arm. By collecting velocity contact force arm operations assembly, analyze change establish active compliance model S-type...
The deep convolutional neural network (CNN) is exploited in this work to conduct the challenging channel estimation for mmWave massive multiple input output (MIMO) systems. inherent sparse features of MIMO channels can be extracted and supports learnt by multi-layer CNN-based through training. Then accurate inference efficiently implemented using trained network. accuracy spectrum efficiency further improved fully utilizing spatial correlation among different antennas. It verified simulation...
Intelligent recognition of electroencephalogram (EEG) signals has been an important means to recognize emotions. Traditional user-independent method, which treat <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">s</sup> each individual's EEG data as independent and identically distributed (i.i.d.) samples ignores destruction on i.i.d. condition caused by individual differences, usually lower generalization performance. Although user-dependent...
Immediate rewards are usually very sparse in the real world, which brings a great challenge to plain learning methods. Inspired by fact that emotional reactions incorporated into computation of subjective value during decision-making humans, an emotion-motivated framework is proposed this article. Specifically, we first build brain-inspired computational model amygdala-hippocampus interaction generate reactions. The intrinsic emotion derives from external reward and episodic memory...
Reinforcement learning recently has achieved impressive success in allowing robots to learn complex motor skills simulation environments. However, most of these successes are difficult transfer physical since current algorithms require lots practical training and sim-to-real skills. To improve the efficiency adaptability robots, this article proposes a guided model-based policy search (GMBPS) algorithm inspired by hypothetical model-free (MF) (MB) actor-critic brain implementation. This...
The inherent value system of a developmental agent enables autonomous mental development to take place right after the agent's "birth." Biologically, it is not clear what basic components constitute system. In computational model introduced here, we propose that systems should have at least three components: punishment, reward and novelty with decreasing weights from first component last. Punishments rewards are temporally sparse but dense. We present biologically inspired architecture...
Heat shock protein 90 (Hsp90) takes part in the developments of several cancers. Novobiocin, a typically C-terminal inhibitor for Hsp90, will probably used as an important anticancer drug future. In this work, we explored valuable information and designed new novobiocin derivatives based on three-dimensional quantitative structure-activity relationship (3D QSAR). The comparative molecular field analysis similarity indices models with high predictive capability were established, their...
This article proposes a novel decision-making framework that bridges gap between model-based (MB) and model-free (MF) control processes through only adjusting the planning horizon. Specifically, output policy is obtained by solving model predictive problem with locally optimal state value as terminal constraints. When horizon decreases to zero, MB will transform into MF smoothly. Meanwhile, inspired neural mechanism of emotion modulation on decision-making, we build biologically plausible...
Histopathological diagnosis of bone tumors is challenging for pathologists. We aim to classify histopathologically in terms aggressiveness using deep learning (DL) and compare performance with pathologists.A total 427 pathological slides were produced scanned as whole slide imaging (WSI). Tumor area WSI was annotated by pathologists cropped into 716,838 image patches 256 × pixels training. After six DL models trained validated patch level, evaluated on testing dataset binary classification...
Purpose Current reinforcement learning (RL) algorithms are facing issues such as low efficiency and poor generalization performance, which significantly limit their practical application in real robots. This paper aims to adopt a hybrid model-based model-free policy search method with multi-timescale value function tuning, aiming allow robots learn complex motion planning skills multi-goal multi-constraint environments few interactions. Design/methodology/approach A goal-conditioned tuning...