I am a fourth year computer science PhD student advised by Professor George Konidaris in the Intelligent Robot Lab at Brown University. I am also fortunate to have done research at Boston Dynamics Robotics and AI Institute.

My research bridges language, perception, and robotics to develop instruction-following systems that improve with experience. I currently focus on methods that equip robots with general-purpose intelligence—enabling them to interpret open-ended instructions, build predictive models, and tackle multiple tasks in unstructured environments.

Verifiably Following Complex Robot Instructions with Foundation Models

Benedict Quartey*, Eric Rosen*, Stefanie Tellex, George Konidaris
NERC '24 (Oral), ICRA 2025
Website  •   ArXiv   •   Poster   •   Thread

TL;DR: A novel approach that enables robots to verifiably follow expressive and complex open-ended mobile manipulation instructions in real-world environments without prebuilt semantic maps.

Bootstrapping Object Level Planning with Large Language Models

David Paulius, Alejandro Agostini, Benedict Quartey, George Konidaris
AAAI Workshop on Generalization in Planning (Spotlight Paper), ICRA 2025
Paper  •  

TL;DR: We use LLMs to extract object-level plan schemas as functional object-oriented networks (FOON), which are then converted into PDDL subgoals for hierarchical task and motion planning.

TaskExplore: Exploiting Contextual Structure to Generate Useful Auxiliary Tasks

Benedict Quartey, Ankit Shah, George Konidaris
NeurIPS Workshop on Generalization in Planning, December 2023
Paper  •   Extended Poster

TL;DR: A framework for multi-task learning that enables off-policy reinforcement learning agents to maximally reuse experience collected while training on single long-horizon tasks.

Affordable Modular Autonomous Vehicle Development Platform

Benedict Quartey, Ayorkor Korsah
IEEE 7th International Conference on Adaptive Science & Technology (ICAST), 2018
Keynote Talk @ 2019 Saint-Étienne International Design Biennale.
Blog  •   Paper  •   IEEE  •   Talk  •   Code  •   Press

TL;DR: RollE is a low-cost, modular platform designed to democratize autonomous vehicle research. It gained traction among African researchers and served as the test platform for a local government-backed climate data collection initiative at the Cité du Design institute in France. In 2019, it was exhibited at the curated Saint-Étienne International Design Biennale