http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Audio-Visual Attention Control of a Pan-Tilt Telepresence Robot
Keng Peng Tee,Rui Yan,Yuanwei Chua,Zhiyong Huan 제어로봇시스템학회 2013 제어로봇시스템학회 국제학술대회 논문집 Vol.2013 No.10
This paper presents an audio-visual attention control system to establish focus of attention for a pan-tilt telepresence robot. Our system is able to detect attentions automatically and reliably in a telepresence scenario using audio-visual stimuli based on speech and face detection. We also describe an audio-visual attention fusion method with context-dependent prioritization among competing modules of face tracking and sound localization. In comparison to the current telepresence system, our system involves automatic tracking of users’ movement and the user need not manually control the robot. We further implemented this attention control system on a pan-tilt robot, and performed an experimental study that illustrates tracking performance when interacting in real time with human users.
Continuous Critic Learning for Robot Control in Physical Human-Robot Interaction
Chen Wang,Yanan Li,Shuzhi Sam Ge,Keng Peng Tee,Tong Heng Lee 제어로봇시스템학회 2013 제어로봇시스템학회 국제학술대회 논문집 Vol.2013 No.10
In this paper, optimal impedance adaptation is investigated for interaction control in constrained motion. The external environment is modeled as a linear system with parameter matrices completely unknown and continuous critic learning is adopted for interaction control. The desired impedance is obtained which leads to an optimal realization of the trajectory tracking and force regulation. As no particular system information is required in the whole process, the proposed interaction control provides a feasible solution to a large number of applications. The validity of the proposed method is verified through simulation studies.
Automated Door Opening Scheme for Non-holonomic Mobile Manipulator
Albertus Hendrawan Adiwahono,Chua Yuanwei,Tee Keng Peng,Liu Bingbing 제어로봇시스템학회 2013 제어로봇시스템학회 국제학술대회 논문집 Vol.2013 No.10
This paper describes a scheme on automated door opening for a non-holonomic mobile manipulator. The perception is done with a fast 2D door plane scan and followed by a 3D door handle recognition. A driving planner is used to drive robot to the intended spot in front of the door with minimum off center error. Then, a simple yet effective downstroke motion of the arm is used to turn the handle. Finally, experiments is used to validate and evaluate the proposed scheme.
Discrete Task-Space Automatic Curriculum Learning for Robotic Grasping
Anil Kurkcu,Cihan Acar,Domenico Campolo,Keng Peng Tee 제어로봇시스템학회 2021 제어로봇시스템학회 국제학술대회 논문집 Vol.2021 No.10
Deep reinforcement learning algorithms struggle in the domain of robotics where data collection is time consuming and in some cases safety-constrained. For sample-efficiency, curriculum learning has shown good results in deep learning-based methods. However, the issue lies on the generation of the curriculum itself, which the field of automatic curriculum learning is trying to solve. We present an automatic curriculum learning algorithm for discrete task-space scenarios. Our curriculum generation is based on difficulty measure between tasks and learning progress metric within a task. We apply our algorithm to a grasp learning problem involving 49 diverse objects. Our results show that a policy trained based on a curriculum is both sample efficient compared to learning from scratch and able to learn tasks that the latter could not learn within a reasonable amount of time.
Model-Based Reinforcement Learning with LSTM Networks for Non-Prehensile Manipulation Planning
Jeffrey Fong,Domenico Campolo,Cihan Acar,Keng Peng Tee 제어로봇시스템학회 2021 제어로봇시스템학회 국제학술대회 논문집 Vol.2021 No.10
Solving non-prehensile manipulation tasks requires domain knowledge involving various interactions such as switching contact dynamics between the robot and the object, and the object-environment interactions. This results in a switched nonlinear dynamic system governing the physical interactions between the object and the environment. In this paper, we propose an interactive learning framework that allows a robot to autonomously learn and model an unknown object’s dynamics, as well as utilise the learned model for efficient planning in completing re-positioning tasks using non-prehensile manipulation. First, we model the overall object dynamics using a Long Short-Term Memory (LSTM) neural network. We then assimilate the learned model into the Monte Carlo Tree Search (MCTS) algorithm with a dense reward function to generate an optimal sequence of push actions for task completion. We demonstrate the framework in both simulated and real robot that pushes objects on a table.