Welcome!

 

I'm a PhD student in the Learning and Intelligent Systems lab at MIT. Here are some projects that I've worked on over the years. Feel free to email me at "jglov at mit dot edu" if you have any questions or would like help using any of my code, or if you just want to say hi!

The Bingham Distribution

 

The Bingham distribution is a type of probability distribution on directional data. I frequently use it in my research to represent uncertainty on 3-D orientations (parameterized by unit quaternions), so I developed an open-source library in C and Matlab, called the Bingham Statistics Library. Please see my research page or one of the papers below for more information.

Bingham Procrustean Alignment for Object Detection in Clutter. Jared Glover and Sanja Popovic. (IROS 2013)

Tracking 3-D Rotations with the Quaternion Bingham Filter. Jared Glover and Leslie Pack Kaelbling. (Technical Report MIT-CSAIL-TR-2013-005, 2013)

Unsupervised Learning for Efficient Texture Estimation From Limited Discrete Orientation Data. Stephen Niezgoda and Jared Glover. (Metallurgical and Materials Transactions A, 2013)

Monte Carlo Pose Estimation with Quaternion Kernels and the Bingham Distribution. Jared Glover, Gary Bradski and Radu Rusu. (RSS 2011)

3D Object Recognition

 

With the recent availability of cheap RGB-D cameras like the Kinect, robust 3-D object recognition (of rigid object instances) is becoming closer to a reality every day. A major focus of my research since 2010 has been on geometric object detection in cluttered RGB-D images, where the Bingham distribution has proven to be a useful tool for processing the orientation information (e.g., surface normals and principal curvatures) in each scene.

Bingham Procrustean Alignment for Object Detection in Clutter. Jared Glover and Sanja Popovic. (arXiv:1304.7399 [cs.CV], 2013)

Monte Carlo Pose Estimation with Quaternion Kernels and the Bingham Distribution. Jared Glover, Gary Bradski and Radu Rusu. (RSS 2011)

Dynamic Manipulation - Ping Pong

 

My current project is to get a robot to play ping pong. More broadly, I'm interested in dynamic manipulation for robots. One finds dynamic manipulation tasks everywhere--chopping vegetables, stirring ingredients in a bowl, flipping hamburgers, moving furniture, playing sports, and so on. Furthermore, many non-dynamic manipulation tasks become dynamic when the robot is pushed to its limits: for example, when we want the robot to move very quickly, or to manipulate very heavy objects.

The Bingham distribution turns out to be very useful for robotic ping pong, because it can be used to create a very robust algorithm for tracking the spin on the ping pong ball.

Tracking 3-D Rotations with the Quaternion Bingham Filter. Jared Glover and Leslie Pack Kaelbling. (Technical Report MIT-CSAIL-TR-2013-005, 2013)

Adaptating to Observable Changes: Lifelong Learning for Dynamic Robots. Jared Glover (Thesis Proposal, MIT CSAIL, 2012)

Paddle Trajectory Optimization for Table Tennis in 2-D. Jared Glover (6.832 Project, 2010)

     

Occluded Shape Inference

 

Another key task in cluttered environments is to estimate what the hidden parts of objects look like. This ability is useful for manipulation, where the robot must reach behind an object to manipulate it. It is also useful for object recognition and localization, particularly for ruling out object configurations which are "geometrically impossible"--i.e. which would cause two or more objects to occupy the same space, or which would cause parts of an object to occupy space which we know to be empty.

Probabilistic Models of Object Geometry with Application to Grasping. Jared Glover, Daniela Rus and Nicholas Roy. (IJRR 2009)

Probabilistic Models of Object Geometry for Grasp Planning. Jared Glover, Daniela Rus and Nicholas Roy. (RSS 2008)

Probabilistic Procrustean Models for Shape Recognition with an Application to Robotic Grasping. Jared Glover (Master's Thesis, MIT CSAIL, 2008)

  

IMP - The Robotic Walker

 

During my undergrad at Carnegie Mellon, I developed a robotic walker named IMP (Intelligent Mobility Platform) as part of the Nursebot project. The walker was able to self-navigate to park itself and return to the user at the touch of a remote. Then, when the user wanted to get up and go, it could provide either audio or visual navigational guidance to the user.

Learning user models of mobility-related activities through instrumented walking aids. Jared Glover, Sebastian Thrun, and Judy Matthews. (ICRA 2004)

A robotically-augmented walker for older adults. J. Glover, D. Holstius, M. Manojlovich, K. Montgomery, A. Powers, J. Wu, S. Kiesler, J. Matthews, and S. Thrun. (Technical Report CMU-CS-03-170, 2003).

  

Carmen

 

I got my start in robotics in 2002 as a developer for Carmen, one of the early open-source robotics libraries.