Research Projects

Contents

Info

I stopped publishing new research in 2018 upon transitioning into an industry role after my PhD. I am still actively involved in organizing and speaking at workshops, serving as a reviewer, associate editor, or program/area chair for conferences and journals, authoring survey papers, and helping with outreach events.

Trajectory Optimization

Idea

Leveraging insights into the structure of the system dynamics or environment allows us to re-formulate nonlinear optimization problems for trajectory planning into friendlier formats (e.g. quadratic/linear/mixed integer programs).

Dynamic constraints

Trajectory optimization for a quadrotor with a suspended payload can be formulated as a Mixed Integer Quadratic Program that still incorporates dynamic, as opposed to simply quasi-static, constraints.

For example, a robot has to “wind-up” and swing the object to carry a payload through a window shorter than the cable length. While seemingly impractical, these types of large swings are actually executed by helicopter pilots during time-sensitive tasks like tree harvesting and firefighting.

Publications

Sarah Tang and Vijay Kumar. “Mixed integer quadratic program trajectory generation for a quadrotor with a cable-suspended payload”. IEEE International Conference on Robotics and Automation (ICRA). Seattle, WA. May 2015.

1
2
3
4
5
6
7
8


@inproceedings{TangKumarICRA2015, 
author = {Sarah Tang and Vijay Kumar}, 
title = {Mixed Integer Quadratic Program Trajectory Generation for a Quadrotor with a Cable-Suspended Payload}, 
booktitle = {2015 IEEE International Conference on Robotics and Automation (ICRA)}, 
year = {2015}, 
pages={2216--2222}, 
doi={10.1109/ICRA.2015.7139492},
}

Scalable coordination

Centralized, multi-robot trajectory optimization in the joint state space quickly grows to an intractable number of decision variables and constraints (e.g. pairwise collision constraints) as team size increases. By using a discretized search step to heuristically allocate space-time “corridors” to robots, the centralized planning problem becomes parallel, decoupled Quadratic Programs.

Publications

Sarah Tang, Justin Thomas, and Vijay Kumar. “Hold Or take Optimal Plan (HOOP): a quadratic programming approach to multi-robot trajectory generation”. The International Journal of Robotics Research (IJRR), vol. 37, no. 9, pp. 1062—1084. Aug. 2018. doi: 10.1177/0278364917741532

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11


@article{TangThomasKumarIJRR2018, 
author = {Sarah Tang and Justin Thomas and Vijay Kumar}, 
title ={Hold Or Take Optimal Plan (HOOP): A Quadratic Programming Approach to Multi-Robot Trajectory Generation},
journal = {The International Journal of Robotics Research},
volume = {37},
number = {9},
pages = {1062--1084},
month = {Aug},
year = {2018},
doi = {10.1177/0278364917741532},
}

Sarah Tang, Justin Thomas, and Vijay Kumar. “Safe navigation of quadrotor teams to labeled goals in limited workspaces”. International Symposium on Experimental Robotics (ISER). Tokyo, Japan. Oct. 2016.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11


@incollection{TangThomasKumarISER2017, 
author = {Sarah Tang and Justin Thomas and Vijay Kumar}, 
title = {Safe Navigation of Quadrotor Teams to Labeled Goals in Limited Workspaces}, 
booktitle = {2016 International Symposium on Experimental Robotics}, 
editor = {Dana Kuli{\'{c}} and Yoshihiko Nakamura and Oussama Khatib and Gentiane Venture}, 
year = {2017}, 
publisher={Springer International Publishing}, 
address={Cham}, 
pages={86--598}, 
doi={10.1007/978-3-319-50115-4_51}, 
}

Sarah Tang and Vijay Kumar. “Safe and complete trajectory generation for robot teams with higher-order dynamics”. IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). Daejeon, Korea. Oct. 2016.

1
2
3
4
5
6
7
8


@inproceedings{TangKumarIROS2016, 
author = {Sarah Tang and Vijay Kumar}, 
title = {Safe and Complete Trajectory Generation for Robot Teams with Higher-Order Dynamics}, 
booktitle = {2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)}, 
year = {2016}, 
pages={1894--1901}, 
doi={10.1109/IROS.2016.7759300},
}

Sarah Tang and Vijay Kumar. “A complete algorithm for generating safe trajectories for multi-robot teams". International Symposium on Robotics Research (ISRR). Sestri Levante, Italy. Sept. 2015.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11


@incollection{TangKumarISRR2015,
author = {Sarah Tang and Vijay Kumar}, 
title = {A Complete Algorithm for Generating Safe Trajectories for Multi-robot Teams}, 
booktitle = {Robotics Research: Volume 2}, 
editor = {Antonio Bicchi and Wolfram Burgard}, 
year = {2018}, 
publisher={Springer International Publishing}, 
address={Cham}, 
pages={599--616}, 
doi={10.1007/978-3-319-60916-4_34}, 
}

Sarah Tang, Koushil Sreenath, and Vijay Kumar. “Multi-robot trajectory generation for an aerial payload delivery system”. International Symposium on Robotics Research (ISRR). Puerto Varas, Chile. Dec. 2017.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11


@incollection{TangSreenathKumarISRR2017,
author = {Sarah Tang and Koushil Sreenath and Vijay Kumar}, 
title = {Multi-Robot Trajectory Generation for an Aerial Payload Transport System}, 
booktitle = {Robotics Research}, 
editor = {Nancy M. Amato and Greg Hager and Shawna Thomas and Miguel Torres-Torriti}, 
year = {2020}, 
publisher={Springer International Publishing}, 
address={Cham}, 
pages={1055--1071}, 
doi={10.1007/978-3-030-28619-4_70}, 
}

Real-time planning

We can execute the complete sensing, occupancy mapping, real-time planning, and trajectory execution pipeline on a quadrotor with only onboard sensing and compute, achieving high-speed flight while maintaining safety, even with respect to unseen obstacles.

Our paradigm uses search-based planning to generate candidate homotopies within the known map, solves a Quadratic Program to generate candidate trajectories (including a contingency maneuver), and selects the most promising action. This breakdown, as opposed to a single global optimization, allows for planning at real-time rates.

Publications

Sikang Liu, Michael Watterson, Sarah Tang, and Vijay Kumar. “High speed navigation for quadrotors with limited onboard sensing”. IEEE International Conference on Robotics and Automation (ICRA). Stockholm, Sweden. May 2016.

1
2
3
4
5
6
7
8


@inproceedings{LiuWattersonTangKumarICRA2016, 
author = {Sikang Liu and Michael Watterson and Sarah Tang and Vijay Kumar}, 
title = {High Speed Navigation for Quadrotors with Limited Onboard Sensing}, 
booktitle = {2016 IEEE International Conference on Robotics and Automation (ICRA)}, 
year = {2016}, 
pages={1484--1491},  
doi={10.1109/ICRA.2016.7487284}, 
}

Controls

Idea

Controllers are generally reliable in nominal operating regimes; we need to achieve equally stable performance in high-dimensional state spaces and at the boundaries of robots' dynamic capabilities.

Geometric control

A promising class of provably almost globally stable geometric controllers for suspended-payload manipulation tasks had previously yet to be used on real robots, partially because idealistic assumptions about the cables' and paylods' dynamic models. Using a downward facing camera and an onboard IMU, we estimate the payload state accurately enough to robustly control agile maneuvers that include payload swings of up to 50 degrees from the vertical.

Some cool stuff

1 of 4 nominees for the IEEE ICRA 2018 Best Paper Award on Unmanned Aerial Vehicles

Publications

Sarah Tang^, Valentin Wüest^, and Vijay Kumar. (^Equal contribution.) “Aggressive flight with suspended payloads using vision-based control”. Robotics and Automation Letters (RA-L), vol. 3, no. 2, pp. 1152—1159, Apr. 2018. doi: 10.1109/LRA.2018.2793305 Presented at IEEE International Conference on Robotics and Automation (ICRA) Brisbane, Australia. May 2018.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12


@article{TangWueestKumarRAL2018, 
author = {Sarah Tang and Valentin W\"{u}est and Vijay Kumar}, 
title = {Aggressive Flight With Suspended Payloads Using Vision-based Control}, 
note={The first two authors contributed equally to this work.}, 
journal = {IEEE Robotics and Automation Letters (RA-L)}, 
volume={3}, 
number={2}, 
pages={1152--1159}, 
month={April}, 
year={2018}, 
doi={10.1109/LRA.2018.2793305},  
}

Formation control

Decentralized PD control laws, when formulated in the appropriate state space, allow Autonomous Underwater Vehicle (AUV) teams to stably track acoustic-tagged targets for long durations, considering factors like maintaining safe following distances, minimizing sensor overlap, arbitrary team sizes, and following multiple target location hypotheses.

We successfully track a leopard shark with a commercial AUV platform. This type of technology could allow biologists to collect more comprehensive data about long migratory aquatic species than (at-the-time) current tracking methods like satellite tags or human tracking.

Publications

Dylan Shinzaki, Chris Gage, Sarah Tang, Mark A. Moline, Barrett Wolfe, Christopher G. Lowe, and Christopher M. Clark. “A multi-AUV system for cooperative tracking and following of leopard sharks". IEEE International Conference on Robotics and Automation (ICRA). Karlsruhe, Germany. May 2013.

1
2
3
4
5
6
7
8


@inproceedings{ShinzakiICRA2013, 
author = {Dylan Shinzaki and Chris Gage and Sarah Tang and Mark A. Moline and Barrett Wolfe and Christopher G. Lowe and Christopher M. Clark}, 
title = {A Multi-AUV System for Cooperative Tracking and Following of Leopard Sharks}, 
booktitle = {2013 IEEE International Conference on Robotics and Automation (ICRA)}, 
year = {2013}, 
pages={4153--4158}, 
doi={10.1109/ICRA.2013.6631163}, 
}

Sarah Tang, Dylan Shinzaki, Chris G. Lowe, and Chris M. Clark. “Multi-robot control for circumnavigation of particle distributions". International Symposium on Distributed Autonomous Robotic Systems (DARS). Baltimore, MD. Nov. 2012.

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11


@incollection{TangShinzakiLoweClarkDARS2012, 
author = {Sarah Tang and Dylan Shinzaki and Chris G. Lowe and Chris M. Clark}, 
title = {Multi-robot Control for Circumnavigation of Particle Distributions}, 
booktitle = {Distributed Autonomous Robotic Systems: The 11th International Symposium}, 
editor = {M. Ani Hsieh and Gregory Chirikjian}, 
year = {2014}, 
publisher={Springer-Verlag Berlin Heidelberg}, 
address={New York}, 
pages={149--162}, 
doi={10.1007/978-3-642-55146-8_11}, 
}

Task Allocation

Idea

Thoughtfully telling which robot what to do makes the whole team more efficient.

Learning-based

In framing multi-robot goal assignment and trajectory planning as a multi-agent reinforcement learning problem, we can develop a general framework applicable to arbitrary robot dynamics.

Publications

Arbaaz Khan, Chi Zhang, Shuo Li, Jiayue Wu, Brent Schlotfeldt, Sarah Tang, Alejandro Ribeiro, Osbert Bastani, and Vijay Kumar. “Learning safe unlabeled multi-robot planning with motion constraints”. IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). Macau. Nov, 2019.

1
2
3
4
5
6
7
8


@inproceedings{KhanIROS2019, 
author = {Arbaaz Khan and Chi Zhang and Shuo Li and Jiayue Wu and Brent Schlotfeldt and Sarah Tang and Alejandro Ribeiro and Osbert Bastani and Vijay Kumar}, 
title={Learning Safe Unlabeled Multi-Robot Planning with Motion Constraints}, 
booktitle={2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)}, 
year={2019},
pages={7558--7565},
doi={10.1109/IROS40897.2019.8968483},
}

Market-based

Leveraging complementary sensor and terrain capabilities helps a heterogenous ground robot team effectively map a rugged tunnel environment.

Five ground robots for cooperative mapping

Publications

Ammar Husain, Heather Jones, Balajee Kannan, Uland Wong, Tiago Pimentel, Sarah Tang, Shreyansh Daftry, Steven Huber, and William L. “Red" Whittaker. “Mapping planetary caves with an autonomous, heterogeneous robot team". IEEE Aerospace Conference. Big Sky, MT. Mar. 2013.

1
2
3
4
5
6
7
8


@inproceedings{HusainIEEE2013, 
author = {Ammar Husain and Heather Jones and Balajee Kannan and Uland Wong and Tiago Pimentel and Sarah Tang and Shreyansh Daftry and Steven Huber and William L. ``Red" Whittaker}, 
title = {Mapping Planetary Caves with an Autonomous, Heterogeneous Robot Team}, 
booktitle = {2013 IEEE Aerospace Conference}, 
year = {2013}, 
pages={1--13}, 
doi={10.1109/AERO.2013.6497363}, 
}