I am currently a PhD candidate at the Robotics Institute at Carnegie Mellon University. My doctoral work is on efficient motion planning for manipulation tasks. I received my B.S. in Electrical and Computer Engineering from Olin College of Engineering in 2008 and my M.S. in Robotics from Carnegie Mellon in 2012.
I am especially interested in applying state-of-the-art planning and control algorithms to interesting real-world robotic systems. In 2008-2009, I worked as a robotics engineer at Barrett Technology, where I wrote a hard-realtime control library for the Whole-Arm Manipulator (WAM) robot. In 2012 I worked as a student on CMU's software entry into the DARPA Autonomous Robotic Manipulation competition. In 2012-2013, I worked as a robotics engineer as a core member of the software team for CHIMP, CMU's entry into the DARPA Robotics Challenge (DRC) competition, in which we placed third out of sixteen teams at the DRC Trials in December 2013.
Here is a copy of my resume.
Feb 2014 - Present
Thesis committee: Siddhartha Srinivasa, Anthony Stentz, Maxim Likhachev, and Lydia Kavraki
My doctoral research focuses on autonomous planning approaches for coupled manipulation tasks in unstructured environments under significant time and energy constraints, motivated by my work on the DARPA Robotics Challenge (DRC). I contend that the problem's cost structure favors explicit handling of both graph representation and task effort optimization, and propose a search algorithm which captures these insights given a model of planning effort. I unify previous work examining configuration space structure of related problems (e.g. in multi-step manipulation) into a general set-theoretic formulation which suggests a planning effort model to be exploited by our roadmap search algorithm, yielding a motion planner which efficiently reuses computation between motion planning queries.
Dellin and Srinivasa, A Unifying Formalism for Shortest Path Problems with Expensive Edge Evaluations via Lazy Best-First Search over Paths with Edge Selectors, ICAPS 2016, London, UK.
Dellin, Efficient Manipulation Task Planning via Reuse-Informed Optimization of Planning Effort, Thesis proposal, April 2015.
Dellin and Srinivasa, A General Technique for Fast Comprehensive Multi-Root Planning on Graphs by Coloring Vertices and Deferring Edges, ICRA 2015, Seattle, WA, USA.
Dellin, Strabala, Haynes, Stager, and Srinivasa, Guided Manipulation Planning at the DARPA Robotics Challenge Trials,'' ISER 2014, Marrakech and Essaouira, Morocco.
Open-source motion planner: lemur (bindings for OMPL, OpenRAVE, and PrPy)
Jan 2013 - Jan 2014
Job Title: Robotics Engineer
Result: Our team placed third out of 16 competitors at the DRC Trials.
After joining the team as a graduate student in Sep 2012, I transitioned to a full-time staff position at the NREC to be involved with both hardware design and analysis of the CHIMP disaster response robot, as well as leading the manipulation and motion planning software architecture and implementation. I performed kinematic reachability and actuator analysis in order to guide the robot's design. I also designed and developed components for operator interface, behavior monitoring, trajectory execution, and the high-frequency manipulator torque control loop. During the DRC Trials competition in Dec 2013, I performed as a robot operator, helping CHIMP claim 16 of 20 available points on manipulation tasks.
Motivated in part by disaster response manipulation tasks, I decided in Feb 2014 to return to finish my doctoral work (above).
Stentz, Herman, Kelly, Meyhofer, Haynes, Stager, Zajac, Bagnell, Brindza, Dellin, George, Gonzalez-Mora, Hyde, Jones, Laverne, Likhachev, Lister, Powers, Ramos, Ray, Rice, Scheifflee, Sidki, Srinivasa, Strabala, Tardif, Valois, Vande Weghe, Wagner, and Wellington, CHIMP, the CMU Highly Intelligent Mobile Platform, Journal of Field Robotics, 2015.
Other links: DARPA website | team website
Jun 2011 - Sep 2012
Developed algorithms for manipulation planning on the HERB robotic platform using constrained trajectory optimization. Lead developer of an open-source implementation of the CHOMP optimizer.
King, Klingensmith, Dellin, Dogar, Velagapudi, Pollard, and Srinivasa, Pregrasp Manipulation as Trajectory Optimization. RSS 2013, Berlin, Germany.
Zucker, Ratliff, Dragan, Pivtoraiko, Klingensmith, Dellin, Bagnell, and Srinivasa, CHOMP: Covariant Hamiltonian Optimization for Motion Planning, IJRR 2013.
Open-source planning library: libcd, orcdchomp
Jun 2011 - Sep 2012
Developed planning and control software for both the SRI and the Carnegie Mellon teams. Designed and implemented synchronized closed-chain torque controller for bimanual manipulation tasks.
Klingensmith, Galluzzo, Dellin, Kazemi, Bagnell, and Pollard, Closed-loop Servoing using Real-time Markerless Arm Tracking. ICRA 2013, Karlsruhe, Germany.
Other links: DARPA website (archive) | DARPA BAA | IEEE Spectrum Article | NREC project website
Jun 2008 - Aug 2009
Job Title: Robotics Engineer
As one of the six engineers at the company, I developed and maintained our robotic software library, wrote embedded software for our motor controllers, performed support for customers at various research universities, and attended domestic and international conferences and trade shows. I developed a new version of Barrett's software library for the WAM (TM) robotic arm (still used today), which featured a substantial rewrite of the hard-realtime control loop, articulated kinematics and inverse dynamics algorithms, joint-space and Cartesian-space torque controllers, and trajectory execution. Designed and implemented demonstrations featuring real-time haptic feedback with real and simulated objects.
Open-source robot control library: libbarrett (2009), (c) Barrett Technology, Inc.