This research aims to improve the ability of a large group of robots to perceive and understand their environment by employing agents with different perceptual skills.
Tracking satellites is an important component of space situational awareness (SSA). However, current ground-based tracking approaches rely on centralized detection and require hours to accurately estimate an orbit. A constellation of low-cost, autonomous cube satellites could provide a fast and robustly decentralized architecture for SSA. We propose distributed particles filters as a method to iteratively refine orbit estimates with low communication bandwidth. We demonstrate the feasibility of this approach by implementing our algorithm in simulation. This simulator can also be used to evaluate the parameter space for future satellite constellation design, as well as test the system's robustness to failures.
Maintaining balance in the face of perpurbations is essential to walking and standing. For my masters thesis, I developed controls for LOPES (LOwer-extremity Powered ExoSkeleton, University of Twente) to assist humans with balance recovery after perturbations, using a combination of feed-forward and feedback control (such as hip torques and a PD controller). We found that even simple, single-joint torques are sufficient to reduce the time to a recovery movement and the energy used by subjects in recovery.
We examined human control of physical interaction with objects that exhibit complex dynamics, hypothesizing that humans exploited stability properties of the human-object interaction. Using a simplified 2D model for carrying a cup of coffee, we developed a virtual implementation to identify human control strategies. The specific task was to transport the cart-pendulum model of a cup of coffee to a target, as fast as possible, while accommodating assistive and resistive perturbations. To assess trajectory stability, we applied contraction analysis. We showed that when the perturbation was assistive, humans absorbed the perturbation by controlling cart trajectories into a contraction region prior to the perturbation. When the perturbation was resistive, subjects passed through a contraction region following the perturbation. Entering a contraction region stabilizes performance and makes the dynamics more predictable. This human control strategy could inspire more robust control strategies for physical interaction in robots.
2018. Stability and Predictability in Dynamically Complex Physical Interactions. In 2018 IEEE International Conference on Robotics and Automation (ICRA), 5540–5545.
2018. Stability and predictability in human control of complex objects. Chaos, 28, 10. DOI: https://doi.org/10.1063/1.5042090
Participants were asked to perform a task in which one arm is moved rhythmically while the other makes fast, discrete movements when cued. Over 20 sessions of practice, participants learned the task asymmetrically: while they learned to make much faster discrete movements, they failed to attenuate the perturbation these discrete movements caused in the rhythmic arm. After 6 months, subjects retained the skills they learned, including the asymmetry.
I proposed this research in my application the Barry Goldwater Scholarship and conducted the work for my undergraduate honors thesis. It was also funded by two Provost Undergraduate Research Awards.
2018. Multi-feature collective decision making robot swarms. In Proceedings of the 17th International Conference on Autonomous Agents and MultiAgent Systems, 1711–1719.
2018. Stability and Predictability in Dynamically Complex Physical Interactions. In 2018 IEEE International Conference on Robotics and Automation (ICRA), 5540–5545.
2018. Stability and predictability in human control of complex objects. Chaos, 28, 10. DOI: https://doi.org/10.1063/1.5042090
2018. Experiments on Human Balancing Co-Control in Standing. Submitted to IEEE RA-L.
Asymmetric Learning in an Asymmetric Bimanual Task. In preparation.
2018. Collaborative Autonomy for Space Situational Awareness. Poster at Lawrence Livermore National Laboratory Summer Student Poster Symposium (8 August 2018). Livermore, CA.
2018. Multi-Feature Collective Decision Making in Robot Swarms. Poster at DOE CSGF Program Review (15–19 July 2018). Washington, DC.
2018. Infotaxis in a Multi-agent Sensor Network. Poster at DOE CSGF Program Review (24–27 July 2017). Washington, DC.
2017. Human Balance Augmentation with Lower Limb Exoskeleton Robot. Poster at RehabWeek 2017 workshop: Towards a next generation of wearable robotic devices for human-oriented assistance and therapy (17 July 2017). London, UK.
2016. Preliminary Study on Assisting Balance Recovery with Lower Limb Exoskeleton. Poster at EuroHaptics 2016 (4–7 July 2016). London, UK.
2016. Dynamic Stability in the Control of Complex Objects. Poster at Biomechanics and Neural Control of Movement 2016 (12–17 June 2016). Mt. Sterling, OH.
2015. Asymmetric Learning in an Asymmetric Bimanual Task. Poster at Society for the Neural Control of Movement 25th Annual Meeting (20–24 April 2015). Charleston, SC.
2015. Why You Don't Spill Your Coffee. Poster at Northeastern University Research, Innovation, and Scholarship Expo (RISE) (9 April 2015). Boston, MA.
2014. Learning and exploration in a novel dimensionality-reduction task. Poster at Society for the Neural Control of Movement 24th Annual Meeting (20–25 April 2014). Amsterdam, NL.
2014. Asymmetric Learning in an Asymmetric Bimanual Task. Poster at Northeastern University Research, Innovation, and Scholarship Expo (RISE) (10 April 2014). Boston, MA.
2013. Asymmetric Learning in an Asymmetric Bimanual Task. Poster at Northeast Undergraduate Research and Development Symposium (2–3 March 2013). Biddeford, ME.
2012. Central Fatigue in Cognitive and Motor Performance. Poster at Northeastern University Research, Innovation, and Scholarship Expo (RISE) (29 March 2012). Boston, MA.
It's cookbook-style recipe website using the Wagtail CMS framework for Django. It was a fun challenge for myself to build the whole thing in a weekend and a week's worth of evenings. I was bad at keeping track of recipes from various cookbooks, websites, and a very poorly organized recipe box. Clearly, the easiest solution to this problem was to build an entire website from scratch.
Find a tasty recipe View on GitHubA satirical fake news website, styled after the New York Times and New Yorker. I made it as a Christmas present (for someone named Owen).
Read fake news View on GitHubThis project combines the static website generation of Jekyll with the web components framework of Google Polymer to create a full-featured material design blog template.
View demo View on GitHub
Suture Score is an application to provide instant, objective feedback on suturing technique to surgical trainees. To prepare for assessments and surgery, students need as much practice as possible. Using innovative image recognition techniques, Suture Score aims to enhance trainees' practice to improve their suturing skills.
Suture Score was created at the NHS-sponsored TELHack hackathon, where it won first place. It's now being tested for further development and use within the NHS.
View website"Paul Erdös Warned Us About Amino Acids, But Look What's Happening Now."
"These Facts About Surface-Wave Tomography Will Change the Way You Look at Deformations in the Space-Time Continuum Forever."
Tired of ridiculous clickbait titles all over your social media feeds? Confuse your friends and amuse yourself with science-themed clickbait headlines.
Like the joke, this very stupid "artificial intelligence" is basically a bunch of "if" statements to fill in templates. That's what you get for a procrastination project in finals week.
Make a headline View on GitHub
VROOM was a final project for an artificial intelligence course. It is a simulated robotic vacuum cleaner designed to explore its environment, classify the types of obstacles it encounters, and plan an efficient path to clean dirt in expected locations. It makes use of many classical AI algorithms such as A* search and naive Bayes classification.
View on GitHubBlurber is a web app to create fake book blurbs. Using the titles and descriptions of the top 300 books in every category on Amazon, this app will generate randomly generate new book blurbs with the power of Markov chains.
Blurber was created with Mustafa Camurcu and Shreyas Mahimkar at HackBeanpot 2015, where it was a finalist.
Try it View on GitHub
A short write-up about our Chaos paper: "Concepts from dynamical systems help to elucidate the subtle physics of how we move."
Read article
I was a guest on a live show of Brains On! (a science podcast for kids) at the Cambridge Science Festival. I showed off my LARVAbots and talked about bioinspired collective robotics.
Listen
"TELHack innovators speak about the inspiration behind their idea, challenges along the way and how their innovation will help improve patient care."
Watch Video
"Julia Ebert, S’15, has won a Marshall Scholarship to pursue a one-year master’s of research in bioengineering at Imperial College London starting in the fall.
"Founded by the British government in 1953 to commemorate the Marshall Plan, the postgraduate scholarship allows up to 40 intellectually distinguished American students to study in the United Kingdom each year."
Read article
"Behavioral neuroscience student and Goldwater Scholar, Julia Ebert, discusses her research."
Watch video
"Julia Ebert, a third-year behavioral neuroscience major, has been awarded a prestigious Barry M. Goldwater Scholarship. The merit-based scholarship is awarded annually to 300 college sophomores and juniors nationwide who are studying science, engineering, or mathematics. Scholarship nominees must demonstrate prior experience in a research setting and submit a proposal to the Foundation Board of Trustees based on work that has advanced their sfield of study and furthered their career objectives."
Read article