SURE–Robotics participants will be involved in diverse robotics research projects. Descriptions of labs and example research projects students will pursue are described below, along with the faculty advisors for each project.
Robot Autonomy and Interactive Learning (RAIL) Lab—Sonia Chernova
The RAIL Lab focuses on the development of robotic systems that operate effectively in complex human environments, adapt to user preferences and learn from user input. Most of our work is motivated by the learning from demonstration problem, which seeks to enable a robot to learn new tasks with the aid of demonstrations or instructions performed by a novice human teacher. The lab is home to a human-scale mobile manipulation platform, a robotic arm, and an apartment living space designed to allow the robots to perform useful household tasks, including cooking, cleaning or hosting a party. Available student projects fall across the entire autonomy pipeline, from perception to reasoning to task execution.
Complex Rheology And Biomechanics (CRAB) Lab—Daniel Goldman
Research in the CRAB Lab focuses on problems that involve complex interaction of matter (physical and biological) with materials that can display solid and fluid properties (like sand, mud, bark, brush). For example, how do organisms like lizards, crabs, and cockroaches cope with locomotion on complex terrestrial substrates (e.g. sand, bark, leaves, and grass)? We seek to discover how biological locomotion on challenging terrain results from the nonlinear, many degree of freedom interaction of the musculoskeletal and nervous systems of organisms with materials with complex physical behavior. The study of novel biological and physical interactions with complex media can lead to the discovery of principles that govern the physics of the media. Our approach is to integrate laboratory and field studies of organism biomechanics with systematic laboratory studies of physics of the substrates, as well as to create mathematical and physical (robot) models of both organism and substrate. Discovery of the principles of locomotion on such materials will enhance robot agility on such substrates.
Adaptive Robotic Manipulation (ARM) Laboratory—Frank L. Hammond III
The ARM Lab focuses on leveraging our knowledge of human and animal motion (locomotion, grasp synergies, robustness to uncertainty and variation), novel embedded sensing and actuation methods (soft sensors, variable stiffness and transformable structures, underactuation), and computational design methods (machine learning, evolutionary optimization) to create robotic devices that boast the versatility and adaptability of biological organisms/manipulators while possessing the precision, strength and speed of man-made machines. Topics include underactuated robotic grasping, kinematically redundant manipulation, teleoperative robotic surgery, polymorphic mobile robots, and wearable devices for human augmentation.
Assistive Robotics at Home, Work, Play—Ayanna Howard
Recent successes in commercial robots appear to foreshadow an explosion of promising robotic applications. Robots will be required to perform tasks in environments that are unsafe for humans, and can help provide lifelong learning, healthcare and successful aging for baby-boomers’ independence. In the Human-Automation Systems Lab, we aim to further robot learning for healthcare applications by developing and validating the core technologies needed to integrate teleoperative-based instruction with other modes of interaction. SURE student will investigate strategies for human interaction with teleoperated assistive robots in home environments. The robots have the potential to improve quality of life for older or disabled adults by allowing caregivers or distant family members to conveniently help an individual with activities of daily living such as meal preparation, administering medications, or performing simple diagnostic procedures. In response to a request for help, a caregiver could remotely take control of the robot to assess the situation and physically lend assistance. The REU participant will contribute to creating his/her own task sequence related to daily living, evaluating the capability of achieving the task sequence with a robot manipulator, and then plugging the results into the main line of work in this research.
Learning for Interactive Agents—Charles Isbell
The Laboratory for Interactive Artificial Intelligence focuses on modeling human beings and their interactions using statistical machine learning techniques to perform activity discovery, modeling and recognition. The project the REU student will work on involves interactive machine learning in which a combination of Learning from Demonstration (LfD) and Reinforcement Learning (RL) methods will be used to build embodied agents that can learn from different users. The domain will involve either a simulation of a grasping robot or an embodied game-playing agent, depending upon the background and interest of the students.
Healthcare Robotics—Charles Kemp
The Healthcare Robotics Lab (HRL) focuses on hands-on research with autonomous mobile manipulators—intelligent mobile robots that physically interact with people and the environment. The lab's research is motivated by the potential for these robots to improve the quality and efficiency of healthcare. HRL's research has enabled robots with a variety of user interfaces to perform assistive manipulation tasks such as retrieving objects, delivering objects, opening doors, turning on lights, and cleaning a person's skin. The lab also works closely with potential end-users, including people with disabilities, nurses, and older adults. HRL is multidisciplinary, so SURE students will have the opportunity to be mentored by Robotics Ph.D. students with backgrounds in computer science, biomedical engineering, and mechanical engineering. SURE students will gain experience in programming robots using Ubuntu, Python and ROS.
Dynamic Adaptive Robotic Technologies (DART) Lab—Ani Mazumdar
The DART Laboratory is focused on the examination of robot mobility. Our research explores the interplay between mechanical design, novel sensing, and intelligent control for aquatic, aerial, and terrestrial systems. Current research focus areas include energy efficiency for mobile systems, agility in uncertain environments, and shared intelligent control. Hands-on development and testing of novel mechatronic approaches is a key part of our work, and students can expect to rapidly develop and test their new ideas. SURE Robotics students will work with physical hardware and computational tools to model, design, and test new physical components, control systems, and algorithms. Through this experience they will develop new skills in software (Matlab, micro controllers), machine design, rapid prototyping, and bench-level experimentation.
Design and Control of an Upper-extremity Rehabilitation Device—Jun Ueda
The goal of the REU projects in the Bio-Robotics and Human Modeling Lab is to develop a compact, robust, haptic interface that can be used in functional MRI (magnetic resonance imaging). Interaction with a haptic device could be used to investigate neuroplasticity after stroke, somatosensory and motor functions, and sympathetic nerve activity during motor task learning. The SURE student will be involved in the design and integration of a rehabilitation robotic device that promotes motor functional recovery in the hemiplegic upper limb.
Construction Robots for Mapping Buildings—Patricio Vela
Dr. Vela’s research lab, the IVALab, focuses on the use of computer vision as the primary sensor in closed-loop systems. Collaborative efforts with faculty in Civil Engineering seek to arrive at a framework for automated mapping and 3D reconstruction of as-built structures using a mobile robot equipped with visual sensors. The advent of the Kinect depth sensor has revolutionized the field of computer vision due to its competitive price and off-the-shelf nature. The IVALab has been exploring how this sensor could be used to rapidly map and reconstruction civil. The REU students will be tasked with retooling existing vision algorithms for use by the Kinect sensor, which provides both image and depth content simultaneously. Comparative evaluation will be made to a pure vision-based approach. Further, students will be employed to understand the limitations and capabilities of the sensor for manipulation tasks in order to combine the structural estimation with automated construction robots. When integrated with Building Information Models, the combined sensing and manipulation creates a closed-loop system for automating construction of structures based on architectural designs and their meta-information (timing, ordering, structural element, etc.).
Automation and Mechatronics—Aaron Young
Dr. Young’s research is focused on developing control systems to improve lower limb prosthetic and exoskeleton systems. His research is aimed at developing clinically translatable research that can be deployed on research and commercial systems in the near future. His group focuses on testing and implementing control systems on robotic assistive devices in patient populations. These devices are evaluated based on the changes to human locomotion biomechanics, energetic cost of walking, and muscle activity changes. The group has active projects on both a robotic knee/ankle prosthesis intended for use with transfemoral amputees and a robotic hip exoskeleton for augmenting locomotion functions. REU participants who work with Dr. Young will work with an interdisciplinary group in robotics, mechanical, electrical and biomedical engineering. They will learn how to conduct human subject experiments and work with clinicians in physical therapy and P&O (Georgia Tech’s Prosthetics and Orthotics Program) to do clinically translatable research. Additionally, they will develop expertise in biological signal processing, mechatronic systems, machine learning, robotics and control.
Control and Navigating Marine Robots—Fumin Zhang
The Lab for Autonomous Mobile Networks focuses on the design and control of marine robots and mobile sensor networks. In our lab, the SURE student will perform research on enhancing existing software, hardware, and algorithms to navigate and control marine robots in realistic marine environment. The robots include both commercial grade vehicles and student-developed vehicles. Depending on specific tasks, students will learn to use Matlab, Labview systems or ROS to develop interesting simulations and demonstrations. Opportunities for field work may also be available.