Honghu Xue

PhD student

University of Luebeck

Biography

Honghu Xue investigates in his doctoral thesis “Sample-efficient Deep Reinforcement Learning”. His methods are applied to navigation tasks. He started his PhD study in March 2019. Prior to that, He received his M.Sc. in Embedded Systems Engineering at Albert-Ludwigs-University of Freiburg with the study focus on Reinforcement Learning, Machine Learning and AI.

Interests

Deep Reinforcement Learning (Sample-efficient Deep RL, Model-based RL, Model-free RL, Automatic Curriculum Learning, Efficient Exploration Strategies, Distributional RL)
Deep Learning (Time Series Prediction, Attention Mechanism, Probabilistic Deep Learning, Generation of High Resolution Images, Video Prediction)

Education

PhD in Artificial Intelligence, 2019/03 - 2023/09
Univeristy of Luebeck
MEng in Embedded Systems Engineering, 2013/10 - 2017/12
Albert-Ludwigs Univerity of Freiburg
Diplom(-Ing.) in Information, Communication and Media Technology (Dual-Degree), 2009/09 - 2013/07
Luebeck University of Applied Sciences
BSc in Electrical Engineering and Automation (Dual-Degree), 2009/09 - 2013/07
East China University of Science and Technology

Publications

Deep Reinforcement Learning for Autonomous Navigation in Intralogistics

Even with several advances in autonomous mobile robots, navigation in a highly dynamic environment still remains a challenge …

Harsh Yadav, Honghu Xue, Yan Rudall, Mohamed Bakr, Benedikt Hein, Elmar Rueckert, Ngoc Thinh Nguyen

Deep Reinforcement Learning for Autonomous Navigation in Intralogistics

Cerebellar transcranial current stimulation – An intraindividual comparison of different techniques

Transcranial current stimulation (tCS) techniques have been shown to induce cortical plasticity. As an important relay in the motor system, the cerebellum is an interesting target for plasticity induction using tCS …

Rebecca Herzog, Till M Berger, Martje G Pauly, Honghu Xue, Elmar Rueckert, Alexander Münchau, Tobias Bäumer, Anne Weissbach

Cerebellar transcranial current stimulation – An intraindividual comparison of different techniques

End-To-End Deep Reinforcement Learning for First-Person Pedestrian Visual Navigation in Urban Environments

We solve a pedestrian visual navigation problem with a first-person view in an urban setting via deep reinforcement learning in an end-to-end manner …

Honghu Xue, Rui Song, Julian Petzold, Benedikt Hein, Heiko Hamann, Elmar Rueckert

End-To-End Deep Reinforcement Learning for First-Person Pedestrian Visual Navigation in Urban Environments

Using Deep Reinforcement Learning with Automatic Curriculum Learning for Mapless Navigation in Intralogistics

We propose a deep reinforcement learning approach for solving a mapless navigation problem in warehouse scenarios. In our approach, an automatic guided vehicle is equipped with two LiDAR sensors and one frontal RGB camera and learns to perform a targeted navigation task …

Honghu Xue, Benedikt Hein, Mohamed Bakr, Georg Schildbach, Bengt Abel, Elmar Rueckert

Using Deep Reinforcement Learning with Automatic Curriculum Learning for Mapless Navigation in Intralogistics

ACNMP: Skill transfer and task extrapolation through learning from demonstration and reinforcement learning via representation sharing

To equip robots with dexterous skills, an effective approach is to first transfer the desired skill via Learning from Demonstration (LfD), then let the robot improve it …

M. Tuluhan Akbulut, Erhan Oztop, M. Yunus Seker, Honghu Xue, Ahmet E. Tekden, Emre Ugur

ACNMP: Skill transfer and task extrapolation through learning from demonstration and reinforcement learning via representation sharing

Using Probabilistic Movement Primitives in Analyzing Human Motion Differences Under Transcranial Current Stimulation

In medical tasks such as human motion analysis, computer-aided auxiliary systems have become the preferred choice for human experts for their high efficiency. However, conventional approaches are typically based on user-defined features such as movement onset times, peak velocities, motion vectors, or frequency domain analyses …

Honghu Xue, Rebecca Herzog, Till M Berger, Tobias Bäumer, Anne Weissbach, Elmar Rueckert

Using Probabilistic Movement Primitives in Analyzing Human Motion Differences Under Transcranial Current Stimulation

Interactive Human–Robot Skill Transfer: A Review of Learning Methods and User Experience

Generalizing the operation of robots in dynamical environments regardless of the task complexity is one of the ultimate goals of robotics researchers. Learning from demonstration approaches supported by transfer learning and user feedback offer a remarkable solution to achieve generalization …

Mehmet Ege Cansev, Honghu Xue, Nils Rottmann, Adna Bliek, Luke E Miller, Elmar Rueckert, Philipp Beckerle

Interactive Human–Robot Skill Transfer: A Review of Learning Methods and User Experience

Parameter Optimization for Loop Closure Detection in Closed Environments

Tuning parameters is crucial for the performance of localization and mapping algorithms. In general, the tuning of the parameters requires expert knowledge and is sensitive to information about the structure of the environment …

Nils Rottmann, Ralf Bruder, Honghu Xue, Achim Schweikard, Elmar Rueckert, Ngoc Thinh Nguyen

Sample-Efficient Covariance Matrix Adaptation Evolutional Strategy via Simulated Rollouts in Neural Networks

In this paper, we propose learning a predictor model that allows simulated rollouts in a rank-based black-box optimizer Covariance Matrix Adaptation Evolutional Strategy (CMA-ES) to achieve higher sample-efficiency. …

Honghu Xue, Sven Böttger, Nils Rottmann, Harit Pandya, Ralf Bruder, Gerdhard Neumann, Achim Schweikard, Elmar Rueckert

Sample-Efficient Covariance Matrix Adaptation Evolutional Strategy via Simulated Rollouts in Neural Networks

Talks

Talk in the 2022 IEEE-RAS International Conference on Humanoid Robots

Remote presentation of the paper “End-To-End Deep Reinforcement Learning for First-Person Pedestrian Visual Navigation in Urban Environments”

Nov 29, 2022 9:00 AM — 10:00 AM Remote Presnetation

Talk in the 2022 IEEE-RAS International Conference on Humanoid Robots

Talk on the "AI-Week" event, Luebeck

Introduction of the current research projects using Deep Reinforcmeent Learning to general audience.

Nov 17, 2021 — University of Luebeck

Talk on the "AI-Week" event, Luebeck

Talk in the "AI Ealy career research Consortium"

Online presentation in the “AI Ealy career research Consortium”, 44th German Conference on Artificial Intelligence

Sep 29, 2021 9:00 AM — 12:00 PM Online Presnetation

Talk in Conference ASPAI 2020

Remote presentation of the paper “Sample-Efficient Covariance Matrix Adaptation Evolutional Strategy via Simulated Rollouts in Neural Networks”

Jul 1, 2020 9:00 AM — 10:00 AM Remote Presnetation

Talk in Conference ASPAI 2020

Talk in the International Workshop of Intelligent Autonomous Learning Systems 2019

Attendance on the International Workshop of Intelligent Autonomous Learning Systems 2019

Aug 4, 2019 — Aug 11, 2019 Kleinwalsertal

Talk in the International Workshop of Intelligent Autonomous Learning Systems 2019

Teaching

Lecture "Probabilistic Machine Learning"

I serve as a teaching assistant for this lecture from 2021 to 2023, and present exercise solutions of theoretical questions and programming tasks.

Sep 27, 2021

Lecture "Probabilistic Machine Learning"

Lecture "Reinforcement Learning"

I serve as the teaching assistant of the lecture “Reinforcement Learning - RO4100” in Univerisity of Luebeck.

Apr 27, 2020

Lecture "Reinforcement Learning"

Projects

Active transfer learning with neural networks through human-robot interactions (TRAIN)

The project receives funding from the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) – No 430054590. I offer support on providing Deep Reinforcement Learning algorithms.

Active transfer learning with neural networks through human-robot interactions (TRAIN)

Analog CMOS Circuit Design

Using Cadence Software to develop a schematic two-stage amplifier satisfying the constraints

Analog CMOS Circuit Design

Very Large-Scale Integration (VLSI)

Using VHDL to develop a UKM-910 processor with basic computation functions.

Very Large-Scale Integration (VLSI)

Mixed-Signal CMOS Circuit Design

Using Cadence Software to design the layout of a two-stage amplifier satisfying the constraints

Mixed-Signal CMOS Circuit Design

Contact

xuehonghu2@gmail.com