Samuel Zapolsky

Homepage for Sam Zapolsky

View My GitHub Profile

Samuel Zapolsky’s Homepage | Positronics Lab @ GWU

GWU 2016

My research is focused on discovering how to get robot behavior in simulation to match reality—and vice versa.

My work in robotics has focused on control and simulation for quadrupedal robots (of which I have constructed several successful physical designs).

I earned my Ph. D. while studying robotics in Evan Drumwright’s Positronics Lab at GWU.

My dissertation focused on discovering what circumstances make robots fail and then using simulted tools to modify a robot or its control system to prevent those failures from happening in the future, leading to a better and more robust robots.

Research Overview:

Research at-a-glance

Inverse Dynamics with Contact Force Prediction

Adaptive Integration for Controlling Speed vs. Accuracy in Multi-Rigid Body Simulation

High-level Quadruped Control (Joystick)

Running with a simulated quadruped

Open source software:

My personal projects:

Pacer: A modular, real-time control library (c++) for legged robot control and locomotion.

When developing Pacer, I strove to deliver the complexity of legged locomotion in an easy to control package, like a remote-controlled car, with the open research problems of foot placement and balancing tackled under-the-hood in easily upgradeable and swappable modules.

Projects I contribute to @ Positronics Lab

Moby: A multi-rigid body simulator (c++)

Ravelin: A linear algebra and dynamics library (c++) for robotics

Publications

Legged robotic locomotion

Bradley Canaday, Samuel Zapolsky, Evan M. Drumwright, “Interactive, Iterative Robot Design”, IEEE International Conference on Robotics and Automation (ICRA). Singapore, May 2017. [Canaday2016]

John Shepherd, Samuel Zapolsky, and Evan M. Drumwright, “Fast multi-body simulations of robots controlled with error feedback”, IEEE International Conference on Simulation and Programming for Autonomous Robots (SIMPAR), 2016. [SZD16.pdf]

Samuel Zapolsky, Evan M. Drumwright, “Particle Traces for Detecting Divergent Robot Behavior”, IEEE-RAS International Conference on Humanoid Robots (Humanoids). Cancun, Mexico, Nov 2016. [Zapolsky2016]

Samuel Zapolsky and Evan M. Drumwright, “Detecting divergent robot behavior with multi-rigid body simulation and random sampling”, The Dynamic Walking Conference (DW 2016), Holly, MI, June 2016. [Zapolsky2016]

Zapolsky, S. & Drumwright, E.,”Inverse dynamics with rigid contact and friction”, Autonomous Robots, 2017, 41, 831-863 [Zapolsky2017]

Samuel Zapolsky and Evan M. Drumwright, “Pacer: Modular, real-time software for legged robot planning and control”, The Dynamic Walking Conference (DW 2015), Columbus, Ohio, July 2015. [Zapolsky2015]

Samuel Zapolsky and Evan M. Drumwright. “Quadratic Programming-Based Inverse Dynamics Control for Legged Robots with Sticking and Slipping Frictional Contacts”. The 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2014), Chicago, IL, Sep 2014. [Zapolsky2014]

Samuel Zapolsky, Evan M. Drumwright, Ioannis Havoutis, Jonas Buchli, and Claudio Semini. “Inverse Dynamics for a Quadruped Robot Locomoting Along Slippery Surfaces”. Proc. 16th Intl. Conf. on Climbing and Walking Robots (CLAWAR), Sydney, Australia, July 2013. [Zapolsky2013]

Robotic simulation

Samuel Zapolsky, Evan M. Drumwright, “Adaptive Integration for Controlling Speed vs. Accuracy in Multi-Rigid Body Simulation”, The 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2015), Hamburg, Germany, Sep 2015. [Zapolsky2015a]

Work history:

Toyota Research Institute
Research Scientist / Roboticist
2017 – Present
Palo Alto, CA

Positronics Lab
George Washington University
Ph.D. Candidate
September 2012 – February 2017
Washington, DC
Advisor: Evan Drumwright

Dynamic Legged Systems Lab
Italian Institute of Technology: Department of Advanced Robotics
Visiting Researcher
July 2012 – Sep 2012
Genoa, Italy
Mentors: Jonas Buchli & Claudio Semini

My Resume

Contact Info

Email: samzapo@gmail.com
LinkedIn