I am a roboticist working to expand how agents can become genuine teammates to support people in human-centered environments.
Currently, I am a Research Engineer/Scientist at
Apple.
I completed my PhD in
Robotics from
Georgia Tech, where I explored how agents can
infer user beliefs,
predict teammate performance, and
support collaboration fluency to enable novel human-AI teaming capabilities. I was advised by
Karen Feigh, and also worked with
Sonia Chernova
and
Harish Ravichandar.
During my studies I had the pleasure of interning with:
- Travelers, researching factual verification of LLM responses to large, unstructured, and inconsistent knowledge bases.
- Gatik AI, improving the behavior prediction module of their autonomous vehicle, tailored to their operational domain.
I also had various research collaborations spanning robot learning from demonstration, generative in-painting, operationalized human-AI teaming, and human-in-the-loop agentic systems.
I spend my free time working on side projects and playing piano.
Here is what I have been up to lately:
Sep'25 Conference Publication at IEEE ICIP 2025
Aug'25 Conference Publication at IEEE RO-MAN 2025
Aug'25 Journal Publication at RA-L 2025
Jun'25 Started working at Apple
May'25 Completed PhD studies at Georgia Tech
Mar'25 Successfully passed my PhD Defense!
May'25 Interned at Travelers
Oct'24 Conference Publication at IEEE/RSJ IROS 2024
Sep'24 Symposium Publication at IEEE ICRA@40 2024
Sep'24 Conference Publication at HFES ASPIRE 2024
Sep'24 Conference Publication at AIAA/IEEE DASC 2024
Feb'24 Successfully passed my PhD Proposal!
Oct'23 Conference Publication at IEEE/RSJ IROS 2023
Aug'23 Conference Publication at IEEE RO-MAN 2023
May'23 Completed MS studies at Georgia Tech
Dec'22 Awarded a Grant from Amazon Consumer Robotics ($80k)
Aug'22 Conference Publication at IEEE RO-MAN 2022
Jul'22 Workshop Publication at IJCAI (Safe RL) 2022
May'22 Interned at Gatik AI
Mar'22 Workshop Publication at ACM/IEEE HRI (MLHRC) 2022
Aug'21 Conference Publication at IEEE RO-MAN 2021
Mar'21 Workshop Publication at ACM/IEEE HRI (WYSD) 2021
Aug'20 Began PhD studies at Georgia Tech
Aug'20 Competed at RoboSub 2020
Jun'20 Interned at NextGen Assistive Technologies
Jun'20 Completed BS studies at UC Riverside
Predicting World Belief States in Dynamic Real-World Environments
Preprint
J. Kolb, A. Garg, N. Warner, K. Feigh
We explore how a robot can infer the belief state of a human in an environment through theory of mind capabilities.
Model Cards for AI Teammates: Comparing Human-AI Team Familiarization Methods for High-Stakes Environments
Conference Proceedings of IEEE RO-MAN '25
R. Bowers, R. Agbeyibor, J. Kolb, K. Feigh
We explore methods for familiarizing users to an autonomous wingmen and their effect on team performance and human factors.
Enabling Controllable, Identity Preserving, Non-Rigid Edits in Human-Centric Images
Conference Proceedings of IEEE ICIP '25
N. Warner, J. Kolb, M. Hahn, J. Huang, I. Essa, V. Birodkar
We condition diffusion in-painting with text captions from image sequences to enable text-controllable pose edits.
Investigating Human-AI Team Fluency in Autonomous Medical Evacuation: A Study of Novice Aviator Cognitive States and HAI Interface Design
Conference Proceedings of AIAA AVIATION '25
S. Doda, R. Agbeyibor, C. Cortes, J. Kolb, J. Magalhaes, K. Feigh
Cognitive workload is estimated from a shared-authority medical evacuation scenario with an autonomous pilot.
Use of Simulated Mental Models and Active Replanning for Human-Robot Interaction
Preprint
J. Ren, A. Swaminathan, J. Kolb, Y. Zhao, S. Coogan, K. Feigh
We propose a framework for intelligent user updates in multi-agent command & control by estimating the user's belief state.
Inferring Belief States in Partially-Observable Human-Robot Teams
Conference Proceedings of IEEE/RSJ IROS '24
J. Kolb, K. Feigh
Presents an architecture for predicting user responses to situation responses in human-robot teams, evaluated on a modified OvercookedAI domain.
Joint Intelligence, Surveillance, and Reconnaissance Mission Collaboration with Autonomous Pilots
Conference Proceedings of HFES ASPIRE '24
R. Agbeyibor, V. Ruia, J. Kolb, K. Feigh
Users participate in a simulated ISR domain with an autonomous pilot, and provide feedback on several modes of autonomy we presented them with.
Towards Safe Collaboration Between Autonomous Pilots and Human Crews for Intelligence, Surveillance, and Reconnaissance
Conference Proceedings of IEEE DASC '24
R. Agbeyibor, V. Ruia, J. Kolb, C. Cortes, T. Mancao, S. Coogan, K. Feigh
Control barrier functions are used to control an autonomous pilot of an ISR aircraft with a human teammate.
Run Time Assurance and Human AI Fluency in Manned Autonomous Intelligence Surveillance and Reconnaissance
Conference Proceedings of the AIAA Aviation '24
R. Agbeyibor, V. Ruia, C. Cortes, J. Kolb, S. Coogan, K. Feigh
Users participate in a simulated ISR domain with an autonomous pilot, we evaluate the team's mission performance and fluency across several operational modes.
Impact of Abstraction Levels of Context Information on AI-Advised Decision Making for an Entry Descent and Landing Task
Conference Proceedings of AIAA SciTech '24
D. Srivastava, J. Kolb, and K. Feigh
A joint human-robot decision making task is structured into a three-step process, improving task performance, user trust, and user situation awareness.
The Effects of Robot Motion on Comfort Dynamics of Novice Users in Close-Proximity Human-Robot Interactions
Conference Proceedings of IEEE/RSJ IROS '23
P. Howell, *J. Kolb, *Y. Liu, and H. Ravichandar
User comfort with various robot motion planning strategies is assessed in a close-proximity collaborative assembly task.
The Effects of Inaccurate Decision-Support Systems on Structured Shared Decision-Making for Human-Robot Teams
Conference Proceedings of IEEE RO-MAN '23
*J. Kolb, *D. Srivastava, K. Feigh
Explores the impact of the type of robot recommendation error on mission performance in a structured shared decision-making domain.
Leveraging Cognitive States in Human-Robot Teaming
Conference Proceedings of IEEE RO-MAN '22, [Best Student Paper Finalist, 3/237]
J. Kolb, H. Ravichandar, and S. Chernova
Measurements of user cognitive states are used to inform a role assignment algorithm and improve teaming performance.
Safe Dexterous Manipulation Using Geometric Boundary Constraints
Workshop Proceedings of IJCAI '22 (Safe RL, non-archival)
A. Jain, J. Kolb, and H. Ravichandar
Instance-specific geometric boundary constraints are used with reinforcement learning algorithms to obtain safe high-dimensional robot hand manipulation.
Evaluating the Effectiveness of Corrective Demonstrations and a Low-Cost Sensor for Dexterous Manipulation
Workshop Proceedings of IEEE/ACM HRI '22 (MLHRC, non-archival)
*A. Jain, *J. Kolb, J. Abbess, and H. Ravichandar
Explores the use of DAGGER-like corrections, and demonstrations from a low-cost pose sensor, to improve robot hand performance at a "pick and place" task.
Predicting Individual Human Performance in Human-Robot Teaming
Conference Proceedings of IEEE RO-MAN '21
J. Kolb, M. Kishore, K. Shaw, H. Ravichandar, and S. Chernova
Through a user study we find that scores from user cognitive skill tests correlate to performance at robot teleoperation tasks.
Partially-Observable OvercookedAI Domain
[Research] Python, PyGame, HTML, JS
A revamped OvercookedAI domain that actually works, supports partial-observability, and has a pretty UI.
MAISR: Multi-Agent ISR Gym for Reinforcement Learning
[Research] Python, PyGame, Gym
A multi-agent aircraft ISR environment, suitable for reinforcement learning and human-robot teaming.
Multi-Agent Robot Operation Scenarios
[Research] Python, Flask, HTML, CSS, ROS
A set of three browser-controlled multi-robot operation scenarios in simulation.
Suite of Cognitive Skill Tests
[Research] Python, Flask, HTML, CSS
A set of three short, simple, online tests to measure user cognitive skills.
