GRASP Lab GRASP 2026 PhD Student and Postdoctoral Researcher Brag Sheet!

Congratulations to all of the 2026 GRASP Laboratory alumni who have graduated or moved on to new positions during the 2025 to 2026 academic year! We’d like to highlight some of their outstanding achievements throughout their time at the University of Pennsylvania.

Anish Bhattacharya

Anish was the first to achieve continuous event-based robot control. He developed methods for future robots to be more agile and able with event-based vision.
Thesis title: Enabling Tight Perception-Action Loops for Autonomous Robots with Event-based Vision
Advised by Dr. Vijay Kumar & Dr. Nikolai Matni

Greg Campbell

Greg was a Visiting Assistant Professor of Mechanical Engineering at Lafayette College in 2025 and presented the last of his PhD work at 2 international conferences. He will be an Assistant Professor of Mechanical Engineering at the University of Scranton in Fall 2026.
Thesis title: Elastomeric Strain Limitation for Design of Soft Pneumatic Actuators
Advised by Dr. Mark Yim

Wei-Hsi Chen

Wei-Hsi works on understanding the underlying design principles governing dynamical and legged robots. He led the development of the Kinegami design algorithm, through which he proved that kinematic chains with arbitrary kinematic complexity can be constructed using tubular modules.
Advised by Dr. Cynthia Sung and Dr. Daniel Koditschek

Victoria Edwards

Victoria received her Ph.D. in August of 2025. She is the recipient of an Outstanding Teaching Award, the John A. Goff Prize, an RSS Pioneer Scholar, and a finalist for the IEEE RAS Technical Committee on Multi-Robot Systems (MRS TC) Best Dissertation Award.
Thesis title: Macroscopic Ensemble Methods for Multi Robot Task Assignment in Dynamic Environments
Advised by Dr. M. Ani Hsieh

Spencer Folk

Spencer has taken a job with Aurora Flight Sciences where they design, build, and fly advanced aircraft and enabling technologies.
Thesis title: Real Time Local Wind Inference for Robust Autonomous Navigation
Advised by Dr. Vijay Kumar and Dr. Mark Yim

Edward Hu

Edward researches reinforcement learning and manipulation, with a focus on acquiring robot behavior policies with minimal sensing requirements. Edward’s research received the Best Paper Award in CoRL’22.
Advised by Dr. Dinesh Jayaraman

Rohit Jena

Rohit’s thesis focuses on Fire ANTs, a Riemannian optimization algorithm that performs medical image registration 300–3000 times faster and at significantly higher resolutions than current standards. This opens up totally new avenues in clinical neuroscience and histology. Some applications, e.g., building atlases of ex vivo brains, would take multiple years with existing methods, and 1-2 days with Fire ANTS.
Advised by Dr. Pratik Chaudhari and Dr. James Gee

Wen Jiang

Wen made fundamental contributions to the next best view problem for view synthesis (FisherRF), localization, and mapping, by discovering an efficient computation of Fisher Information in rendering.
Thesis title: Active Perception for 3D Scene Representations from an Information Theoretic Perspective
Advised by Dr. Kostas Daniilidis

Pratik Kunapuli

Pratik developed an efficient reinforcement learning framework for aerial vehicles, enabling agile trajectory tracking of aerial manipulators.
Thesis Title: Advancing Aerial Agility with Reinforcement Learning
Advised by Dr. Dinesh Jayaraman and Dr. Vijay Kumar

Jiahui Lei

Jiahui is a recipient of the 2026 Morris and Dorothy Rubinoff Award awarded by the Computer and Information Science Department for the completion of a doctoral dissertation that represents an innovative application of computer technology. Jiahui made fundamental contributions to dynamic 3D scene reconstruction by introducing novel 4D representations and inference-time optimization.
Thesis title: 4D Vision: Represent, Reconstruct and Generate the Dynamic 3D World
Advised by Dr. Kostas Daniilidis

Alice Kate Li

Alice developed a Koopman framework to learn flow fields, enabling both active environmental sampling and the synthesis of reactive obstacle-aware dynamical systems from demonstrations.
Thesis title: Structured Learning of Flow Fields: From Fluid Modeling to Reactive Robot Navigation
Advised by Dr. Vijay Kumar and Dr. M. Ani Hsieh

Hancheng Min

Hancheng, a former postdoc at ViDAL lab, is now a tenure-track associate professor at Institute of Natural Sciences, Shanghai Jiao Tong University. His research is at the intersection of dynamical systems and machine learning, and the theoretical understanding of training dynamics of neural networks. He received the AI x Science Postdoctoral Fellowship from DDDI and IDEAS at Penn and received the MINDS Data Science Fellowship twice during his Ph.D. at Johns Hopkins University.
Advised by Dr. Rene Vidal

Farhad Nawaz

Farhad’s research focuses on safe and stable hierarchical planning for robots operating in human-centric environments, spanning robot manipulation at Penn and autonomous parking systems in collaboration with the Honda Research Institute. His work has been accepted at multiple prestigious international robotics conferences. Following graduation, Farhad will join Lila Sciences as a Research Scientist working on dexterous manipulation and robot learning.
Thesis title: Task and Motion Plans for Robots in Human-Centric Environments
Advised by Dr. Nikolai Matni and Dr. Nadia Figueroa

Alex Nguyen-Le

Alex explored the role of classical ideas from control theory (Riccati equations) in non-convex optimization, with applications to structured covariance estimation and domain-randomized LQR control.
Thesis title: On Riccati Equations in Nonconvex Optimization
Advised by Dr. Nikolai Matni

Stefanos Pertigkiozoglou

Stefanos introduced fundamental methods for discovering symmetry in data, designing equivariant attention architectures, and novel methods for dealing with approximate equivariance.
Thesis title: Data-Driven and Constrained Approaches to Learning and Enforcing Symmetry in Neural Networks
Advised by Dr. Kostas Daniilidis

Jianing “Aurora” Qian

Aurora’s research lies at the intersection of robot learning, computer vision, and machine learning. She has studied how structured and hierarchical visual representations can be used to build scalable world models and visuomotor policies for long-horizon robotic manipulation and planning. Her work spans object-centric scene representations, hierarchical visuomotor control, and scene graph-based imitation learning.
Tentative thesis title: “Structured Representations for Long-Horizon Robot Learning.”
Advised by Dr. Dinesh Jayaraman

Junyao Shi

Junyao explores how human data and foundation models trained at scale can help overcome bottlenecks, including leveraging in-the-wild human videos to transfer manipulation skills, and providing rich signals for robot learning, and showing how VLMs can automate traditionally hand-engineered components of robotics pipelines, including modular system construction, data collection, reward design, and simulation construction.
Thesis direction: Unlocking Generalist Robots with Human Data and Foundation Models
Advised by Dr. Dinesh Jayaraman

Anusha Srikanthan

Anusha was president of ESE PhD Association (2022-2024), received the PhD good citizen award (2023), and the excellent paper award at the IROS Workshop on Cognitive and Social Aspects of Human Robot Social Interaction.
Thesis title: Layered Frameworks for Robot Planning and Safety: From Conservatism to Performance
Advised by Dr. Vijay Kumar and Dr. Nikolai Matni

Yuezhan Tao

Yuezhan developed a unified framework for active perception that enables efficient, long-horizon, task-driven autonomy in real-world environments.
Thesis title: Robot Planning for Active Perception
Advised by Dr. Vijay Kumar

Yufu Wang

Yufu developed pioneering methods for 3D human reconstruction from monocular video, culminating in TRAM, a landmark system that pushed the boundaries of human mesh and motion recovery in unconstrained real-world scenes.
Thesis title: Grounding and Generating 4D Humans in the Open World
Advised by Dr. Kostas Daniilidis

Ziyun (Claude) Wang

Claude made fundamental contributions to event-based perception. He
introduced self-supervised learning for independent motion detection
and devised algorithms for 3D shape from apparent event contours, 3D
human pose from events, and high-speed interception.
Thesis title: Beyond Frames: Learning to Perceive With Event-Based Vision
Advised by Dr. Kostas Daniilidis

Yinshuang Xu

Yinshuang made fundamental contributions to equivariant deep learning on homogeneous spaces. She was the first to model convolution and attention on light-fields expressed as functions on rays.
Thesis title: Equivariant Learning in 3D Vision
Advised by Dr. Kostas Daniilidis

Fengjun Yang

Fengjun’s PhD research developed structured data-driven control methods for large-scale cyber-physical systems, using locality, graph symmetry, and differential flatness to make learning-based controllers scalable, efficient, and analyzable. His work includes data-driven distributed MPC, graph-structured control, and hardware-validated residual learning for close-proximity multi-quadrotor flight.
Thesis title: Synthesizing Scalable Distributed Controllers by Exploiting and Learning Structure
Advised by Dr. Nikolai Matni

Thomas Zhang

Thomas’ PhD work has surrounded making feedback work for AI systems, including: 1. presenting the fundamental difficulty of imitation learning and providing simple-but-crucial interventions, 2. exposing how distribution shift from feedback exposes pathologies in neural network feature learning, 3. proposing novel optimizer paradigms that fix these pathologies to enable long-horizon performance. His work has been published ACC, CDC, TAC, NeurIPS, ICML, and ICLR (spotlight award).
Tentative thesis title: Learning to Control: Reduction, Learning, Optimization
Advised by Dr. Nikolai Matni