ELPIS Lab

UH200B, Unity Hall

100 Institute Rd

Worcester, MA 01609

Welcome to the webpage of the Efficient Learning and Planning for Intelligent Systems (ELPIS) Lab.

The Lab has a broad interest in autonomous robotic systems capable of reasoning about and interacting with the physical world. The primary goal is to develop agents that are efficient, robust, and capable of learning from real-world interactions. Current research projects focus on the integration of classical planning algorithms and state-of-the-art machine learning techniques, aiming to advance 1) planning efficiency, 2) planning robustness, and 3) planning from visual inputs.

If you are interested in joining the Lab please see this page.

Research Focus Areas

Our core work spans three complementary themes that shape the lab's research agenda.

Research Focus 1

Planning Robustness

We build robust planners that reason under uncertainty and partial observability to improve safety and reliability.

Explore Robustness Research

Planning Robustness

Research Focus 2

Planning Efficiency

We design planning systems that reduce computation time while retaining high-quality trajectories in high-dimensional spaces.

Explore Efficiency Research

Planning Efficiency

Research Focus 3

Planning from Visual Inputs

We investigate representations that enable task and motion planning directly from image observations in real-world scenes.

Explore Vision Research

Planning from Visual Inputs

News

May 30, 2026	Our paper ActivePusher: Active Learning and Planning with Residual Physics for Nonprehensile Manipulation lead by Zhuoyun Zhong, and Ali Golestaneh was accepted to ICRA2026 and is nominated for best paper award in the Planning and Control Category!
May 07, 2026	The senior capstone project (MQP) on parachute packing from ELPIS lab, won the RBE department award! See the amazing video here.
Aug 29, 2025	ELPIS Lab along with professor Jing Xiao recieved an Army Research Center grant for 170k$ to conduct research on real-time path planning in unmapped rough off-road terrains.
Jul 01, 2025	ELPIS Lab was awarded the NSF-CRII grant for 175k$ to conduct research on how robots can learn abstract, symbolic representations from visual and execution data.
Jun 21, 2025	ELPIS lab was at RSS2025 with one talk on MotionBenchMaker and a workshop paper on active learning for non-prehenslie manipulation.
May 30, 2025	ELPIS Lab along with professor Jing Xiao recieved an Amazon WPI Robotics Enginnering Award gift for 75k$ to conduct research on whole-body planning for a humanoid robot.
May 15, 2025	The first cohort of 8 M.S. students from the ELPIS Lab has graduated. Congratulations to everybody!
Mar 15, 2025	Prof. Chamzas was an invited panelist for the Future Faculty Fellows at Rice University.
Jan 24, 2025	Our paper “Multi-layer Motion Planning with Kinodynamic and Spatio-Temporal Constraints” has been accepted to ACM Hybrid Systems: Computation and Control (HSCC) 2025! Congratulations to Abhiroop Ajith and Jeel Chatrola!
Nov 14, 2024	Prof. Chamzas gave a talk at the UNH Robotics Seminar on “Learning and Planning for Robotic Manipulation: Bringing Theory to Practice.”