I'm a Senior Systems Scientist in the Robotics
Institute at Carnegie
Mellon University. I'm affiliated with the National
Robotics Engineering Center, or NREC for short. Additionally, I have a split assignment
with the Computer Science Department at
Carnegie Mellon's Qatar branch campus,
where I co-direct the Qri8 robotics lab.
My main research interests focus on perception and autonomy, although I also have done lots of work
in multirobot and human-robot teams particularly in the robot soccer
domain. I am deeply interested in developing robots that have an impact on society. In particular,
I am very interested in deploying robots to industrial settings where the need for robots to take on
dirty, dull, and dangerous tasks can have a big impact on improving safety and productivity.
I am particularly interested in bringing robots into the Oil and Gas industry and solving
the perception and autonomy challenges faced by robots working in a difficult, semi-structured environment
with or without human operators. I run projects at RI, NREC, and Carnegie Mellon Qatar that
are attempting to address many of these issues.
As I shift between RI on main campus, NREC, and Qatar, the most reliable way to contact me is via email ("brettb (at) cs.cmu.edu").
You can also reach me :
US Cell: +1 412 721 6889
Qatar Cell: +974 551 6924
Carnegie Mellon University
Newell Simon Hall 4000B
5000 Forbes Avenue
Pittsburgh, PA, 15213, USA
Teaching & Activities
Below are the classes I have created, taught, and/or co-taught. Q indicates the class was in Qatar
- 15-381Q: Artificial Intelligence (2010, 2008)
- 16-362Q: Mobile Robot Programming (2009)
- 16-200Q: Introduction to Autonomous Robots (2007, 2005)
- 16-100Q: Introduction to Mobile Robot Programming (2007)
- 15-491: CMRoboBits (2006)
The following are my current or past activities:
Labs and Students
I co-direct the Qri8 robotics lab in Qatar, and am affiliated with the
National Robotics Engineering Center in Pittsburgh. I also work with
Tony Stentz and Bernardine Dias's rCommerce lab
in RI, and have previously worked with Manuela Veloso's Coral research lab.
The following is a list of postdocs and students that I am advising or have advised:
- Dr. Peter Hansen: Postdoc at CMUQ
Hatem Alismai: RI MS and 2009 CMUQ CS alum
- Samreen Anjum: 2010/2011 CMUQ CS Senior Thesis
- Dr. Brenna Argall: RI PhD. Co-advised with Manuela Veloso. (Currently a postdoc at EPFL)
- Kristina Rohlin: RI MS. Co-advised with Manuela Veloso
Searock: RI MS. Co-advised with Manuela Veloso
Govindaraju: RI MS. Co-advised with Manuela Veloso
- Jennifer Lin: CS Senior Thesis
My recent publications are online. Please contact me if you have any questions or any
papers are missing.
The following is a list of my current active projects.
Visual SLAM for Industrial Robots
We are exploring algorithms to support visual mapping and localization for a robot vehicle operating in an industrial setting
such as an LNG production plant. This will be work with Joe Cheben, Tony Stentz, and Bernardine Dias. This work is sponsored by QNRF.
LNG Pipe Vision
We are developing a perception system to fit onto a pipe crawler. Our system will build
3D appearance models of the inside pipe surface and register changes in appearance and structure over time
in order to support automated corrosion detection. This work is with Peter Rander and is sponsored by QNRF.
We are exploring a mix of physics-based and data driven high fidelity sensor modeling techniques. The goal is to develop
a system that can provide much more realistic UGV simulation than current techniques. Such simulation will play a crucial
role in speeding up the development cycle, and in validating platforms. This is work with Peter Rander and is sponsored by
the US ACE ERDC.
We are exploring human robot interaction in a mixed cultural setting. To do this exploration we have developed a robot Hala that runs
in the Carnegie Mellon Qatar central reception and speaks both English and Arabic. This is work with Majd Sakr and Reid Simmons
and is sponsored by QNRF.
These are older projects that are either no longer active, or I am no longer directly involved with.
Learning by Demonstration with Advice
We explored a range of techniques to enable rapid acquisition of skills (ie. control policies) by a robot from a
human teacher. This was work with Brenna Argall and
We developed a distributed coordination architecture to enable dynamically formed teams from
a mixed team of humans and heterogeneous robots to perform tasks. Our approach is
robot architecture agnostic. This was work with Bernardine Dias,
Veloso, Tony Stentz
Alex Rudnicky and was sponsored by Boeing. Please see the
project page for more information.
This project used Segway RMP robots to play
soccer both indoors and outside with human teammates. Research issues are single robot autonomy,
localization, multi-object tracking, skill learning, robust vision,
teamwork with pickup teams. This was work with Manuela
Veloso and was sponsored by DARPA. See our publications, images,
for more information.
We developed a small-size robot soccer team to explore multi-robot coordination in an adversarial environment.
Research issues are team autonomy, opponent modelling, fast path planning and control, robust hardware, skill
learning. This was work with Manuela
Veloso and was sponsored by DARPA. Please see our code, publications, images,
for more information.
Teams of UAVs (ACO)
This program looked at developing coordination algorithms and real-time kinodynamically
constrained path planning for a team of UAVs. This was work with
Veloso and James Bruce and
Aiding Driver Safety
We developed probabilistic algorithms to predict intended driver route based on historical data
using low quality GPS. Additionally, we developed an efficient algorithm to estimate road grade from
low quality 3D GPS fixes. This was work with Reid
Simmons that was sponsored by General Motors.
We developed a vision-based system to observe and infer the interactions at meetings.
This was work with Manuela Veloso, Paul Rybski, Fernando Del Torre,
and Takeo Kanade and was sponsored by DARPA.
We developed a real-time simulation engine to enable rapid prototyping of robot control algorithms with
a modest level of realistic dynamics. Our approach allowed for configurable hardware platforms.
See ubersim for details. This was work with
Manuela Veloso and was extended by
Mike Bowling at the
University of Alberta.
Biologically Plausible Spatial Navigation
My dissertation work focused on developing a vision-based robot
navigation system that was inspired from biologically models of the rat hippocampus. This work
was been continued at the University of Queensland as part of the RatSLAM
project and they now have some very impressive results.