Andrew Kimmel

About

Andrew Kimmel is a robotics researcher whose work spans robot manipulation, motion planning, and data-efficient learning for autonomous systems. His research addresses some of the most demanding challenges in robotics: enabling robots to operate effectively in cluttered, real-world environments with minimal prior knowledge. Kimmel's most influential contributions lie in model identification and learning for robotic manipulation. His work on physics engine-based model identification — garnering over 40 citations — introduced a Bayesian optimization approach that dramatically reduces the real-world experiments needed to estimate mechanical parameters like mass and friction, bridging the sim-to-real gap efficiently. Complementing this, his research on learning stochastic transition models for underactuated robotic hands (36 citations) offers a practical pathway toward deploying low-cost, adaptive grippers without requiring complex analytical models. His contributions to rearrangement planning using pebble graphs (38 citations) demonstrate sophisticated algorithmic thinking for multi-body manipulation in cluttered spaces. Additional work on multi-agent coordination, belief-space planning, and the PRACSYS motion planning architecture reflects the breadth of his systems-level thinking. Collectively, Kimmel's research advances the practical deployment of intelligent robotic systems across manipulation, planning, and learning domains, making him a valuable voice in modern robotics research.

Research Focus

Key Achievements

Top Papers

1. 1
   Fast Model Identification via Physics Engines for Data-Efficient Policy Search

   41 citations · 2018
2. 2
   Rearranging similar objects with a manipulator using pebble graphs

   38 citations · 2014
3. 3
   Learning a State Transition Model of an Underactuated Adaptive Hand

   36 citations · 2019
4. 4
   Maintaining team coherence under the velocity obstacle framework

   17 citations · 2012
5. 5
   PRACSYS: An Extensible Architecture for Composing Motion Controllers and Planners

   17 citations · 2012
6. 6
   Fast, Anytime Motion Planning for Prehensile Manipulation in Clutter

   16 citations · 2018
7. 7
   Evaluating end-effector modalities for warehouse picking: A vacuum gripper vs a 3-finger underactuated hand

   13 citations · 2016
8. 8
   An Extensible Software Architecture for Composing Motion and Task Planners

   11 citations · 2014
9. 9
   Belief-Space Planning Using Learned Models with Application to Underactuated Hands

   11 citations · 2022
10. 10
    Model Identification via Physics Engines for Improved Policy Search.

    10 citations · 2017

Key Collaborators

Contact & Links