WEI DU
Homepage: https://ddwwee.com Email: adcp2f@r.postjobfree.com Tel: 412-***-****
EDUCATION AND ACADEMIC POSITION
Carnegie Mellon University (CMU) 05/2019 - Present
- Research Associate at the Robotics Institute
Carnegie Mellon University (CMU) 05/2019
- Master of Science in Mechanical Engineering (GPA: 3:89=4:0) Harbin Institute of Technology (HIT) 06/2017
- Bachelor of Engineering in Mechanical Design, Manufacturing and Automation (GPA: 3:83=4:00) RESEARCH EXPERIENCE
MHI Project – Search-based Planning Lab (CMU & Mitsubishi Heavy Industries) 01/2020 - Present
- Research on Multi-UAV surveillance and deconfliction problem.
- Applied Multi-Resolution A* algorithm on UAV for large area coverage tasks. Multi-Resolution A* algorithm – Search-based Planning Lab (CMU) 06/2019 - Present
- Researched on executing multiple searches simultaneously to explore search spaces with different resolutions.
- Exploited different scheduling policies such as DTS to speed up the search.
- Implemented MRA* on PR2 robot with significant improved planning performance. Walker Project – Search-based Planning Lab (CMU & UBTech Inc.) 04/2019 - Present
- Applied ARA* algorithm on Walker robot for manipulation tasks on conveyor.
- Constructed perception-planning-grasp pipeline for grasping tasks, e.g., grasp a box from the conveyor. Cruzr Project – Search-based Planning Lab (CMU & UBTech Inc.) 10/2018 - 04/2019
- Customized SLAM algorithms for Cruzr robot to assist its indoor navigation.
- Constructed a state machine to coordinate planning and SLAM.
- Constructed local controllers to execute plans returned from the planner. Planning Using Soft Duplicate Detection – Search-based Planning Lab (CMU) 12/2017 - 02/2019
- Explored search-based planning approaches in continuous state space with soft duplicate detection scheme.
- Implemented decision trees in prioritizing states and map pattern recognition. COURSE PROJECTS (CMU)
Power plant Substation-to-feeder Path Prediction – Bayesian Machine Learning 02/2019 - 05/2019
- Collaborated with Kevala company on predicting feeder-path and outperformed the peer team at Facebook.
- Implemented CNNs to predict the endpoints of power lines with accuracy 92%.
- Employed motion planning algorithms in generating the substation-to-feeder paths. Trajectory Optimization for Fixed-wing Airplane – Egineering Optimization 02/2019 - 05/2019
- Constructed a complex optimization model for NASA’s X 57 Electric Research Plane.
- Applied Differential Dynamic Programming algorithm to solve optimization problem.
- Modeled this problem as a graph search problem and solved this problem with respect to this perspective. Offline Hand-written Chinese Characters Recognition – Pattern Recognition Theory 09/2018 - 12/2018
- Exploited CNNs in recognizing hand-written Chinese characters with accuracy 98%.
- Implemented decision-trees, SVM as baselines against CNNs in recognizing hand-written characters. PUBLICATIONS
Wei Du, Fahad Islam and Maxim Likhachev. Multi-Resolution A*. Proceedings of the International Symposium on Combinato- rial Search (SoCS), 2020.
Wei Du, Sung K. Kim, Oren Salzman and Maxim Likhachev. Escaping Local Minima in Search-Based Planning using Soft Du- plicate Detection, Proceedings of the IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2019 RELATED COURSEWORKS
Datastructures and Algorithms for Engineers; Java for Application Programmers; Robot Localization and Mapping; Planning Techniques for Robotics; Pattern Recognition Theory; Bayesian Machine Learning for Scientists and Engineers; SKILLS
- Programming Language: C/C++, Java, Python, MATLAB;
- Software: ROS, OMPL, SMPL, SBPL, AutoCAD, Solidworks, ADAMS