Homepage - Yi Zheng

Yi Zheng

PhD Student in Computer Science
University of Southern California

About Me

I am a Ph.D. student in Computer Science at the University of Southern California. My research focuses on designing scalable algorithms for complex planning, optimization, and resource-allocation problems.

My current work develops efficient and scalable algorithms for Multi-Agent Path Finding and Virtual Network Embedding. These problems involve coordinating multiple agents, allocating limited resources, and finding feasible paths under complex constraints. My recent research applies AI, search, and optimization techniques to large-scale planning, coordination, and resource allocation problems. I am also interested in translating algorithmic research into practical systems for robotics, autonomous agents, and large-scale network optimization.

Research interests:
Artificial Intelligence · Search & Planning · Multi-Agent Systems · Motion Planning · Combinatorial Optimization · Resource Allocation

Education

University of Southern California

Ph.D. in Computer Science

2019 - Present

Advisors: Prof. Sven Koenig and Prof. T. K. Satish Kumar
University of Southampton

B.S. in Computer Science - 1st class honored

2015 - 2019

Experience

University of California, Irvine

Junior Specialist (Summer Internship Position)

Jun 2025 – Aug 2025
- Designed and developed Multi-Agent Path Finding algorithms for 3D warehouse robots.
- Implemented algorithms in C++ and built simulations with ROS and Gazebo independently.
University of Southern California
Research Assistant

Aug 2019 – Present
- Conduct research on scalable algorithms for Multi-Agent Path Finding and Virtual Network Embedding, developing new heuristic-search and optimization techniques for large-scale combinatorial problems.
- Designed and developed Virtual Network Embedding algorithms that significantly outperform the existing work in scalability and runtime while maintaining near-optimal solution costs.
- Collaborated with the USC Information Sciences Institute (ISI) research team, and my Virtual Network Embedding algorithms were applied to network slicing in the SABRES project, connecting fundamental algorithmic research with practical networking systems.
- Collaborated with Prof. Ken Nakagaki on Human-Computer Interaction and robotics research, developing motion-planning algorithms for the robots.
Teaching Assistant
- DSCI 553 Foundations and Applications of Data MiningJan 2026 – May 2026
- CSCI 567 Machine LearningAug 2025 – Dec 2025
- CSCI 561 Foundations of Artificial IntelligenceJan 2025 – May 2025
- CSCI 561 Foundations of Artificial IntelligenceAug 2024 – Dec 2024
- CSCI 360 Introduction to Artificial IntelligenceAug 2021 – Dec 2021

Projects & Activities

NeurIPS 2020 Flatland Challenge 1st Place

Won 1st place in both rounds of the NeurIPS 2020 Flatland Challenge, a large-scale railway planning and scheduling competition held in partnership with major European railway companies.
Outperformed all other submissions in both tracks, including reinforcement-learning entries, in a field of more than 700 participants from 51 countries and over 2,000 submissions.
The USC press release

AI Education

Contributed to an NSF-supported project developing an advanced manufacturing curriculum by creating slides, videos, exercises, and interactive Python notebooks for AI in manufacturing (with Prof. Sven Koenig).
Contributed to the writing and editing of the chapter Artificial Intelligence and Automation in the Springer Handbook of Automation, 2nd ed. [book]

Publications

Adapting the Conflict-Based Search Framework for the Virtual Network Embedding Problem

Yi Zheng, Erik Kline, Lincoln Thurlow, Srivatsan Ravi, Sven Koenig, T. K. Satish Kumar

Journal of Artificial Intelligence Research (JAIR) Volume 86, Article 10, 2026

Warning

Action required

Education

Experience

Projects & Activities

Publications

Adapting the Conflict-Based Search Framework for the Virtual Network Embedding Problem

Adapting the Conflict-Based Search Framework for the Virtual Network Embedding Problem

Cube-Based Automated Storage and Retrieval Systems: A Multi-Agent Path Finding Approach

Cube-Based Automated Storage and Retrieval Systems: A Multi-Agent Path Finding Approach

Enhancing Lifelong Multi-Agent Path Finding with Cache Mechanism

Enhancing Lifelong Multi-Agent Path Finding with Cache Mechanism

Virtual Network Embedding as Boolean Satisfiability

Virtual Network Embedding as Boolean Satisfiability

Threading Space: Kinetic Sculpture Exploring Spatial Interaction Using Threads In Motion

Threading Space: Kinetic Sculpture Exploring Spatial Interaction Using Threads In Motion

Priority-Based Search for the Virtual Network Embedding Problem

Priority-Based Search for the Virtual Network Embedding Problem

Improved Conflict-Based Search for the Virtual Network Embedding Problem

Improved Conflict-Based Search for the Virtual Network Embedding Problem

Conflict-Based Search for the Virtual Network Embedding Problem

Conflict-Based Search for the Virtual Network Embedding Problem

(Dis)Appearables: A Concept and Method for Actuated Tangible UIs to Appear and Disappear based on Stages

(Dis)Appearables: A Concept and Method for Actuated Tangible UIs to Appear and Disappear based on Stages

Artificial Intelligence and Automation

Artificial Intelligence and Automation

Flatland Competition 2020: MAPF and MARL for Efficient Train Coordination on a Grid World

Flatland Competition 2020: MAPF and MARL for Efficient Train Coordination on a Grid World

Scalable Rail Planning and Replanning: Winning the 2020 Flatland Challenge

Scalable Rail Planning and Replanning: Winning the 2020 Flatland Challenge

Academic Service

Program Committee

Reviewer for Conferences and Journals