Background Image for Header:
NSF REU Site
Undergraduate Robotics Research for Rural Appalachia
This NSF-Funded REU site will promote STEM opportunities through robotics-related research, education, and outreach. Undergraduate students will work in small groups in one of five robotics research labs at WVU under the mentorship of a faculty and perform research in the areas of motion planning, localization, coordination, autonomy, and locomotion for challenging field applications. The proposed research themes are tailored to problems associated with enabling the use of mobile robotics in rural environments that have challenging topography and/or are dense forested areas, typical to the Appalachian region. Participants will perform fundamental research leading to hands-on experimental demonstrations.
Our educational objective is to provide opportunities for undergraduates to conduct independent research for problems that have great societal impact. The students will participate in weekly group research presentations, organized panel discussions, university sponsored workshops, team building activities, and a university-wide undergraduate research symposium. Participants will contribute to the state of the art within a specialized area of robotics, and faculty mentors will use group discussions to emphasize a holistic view of robotics systems and identify potential synergies and collaborations across sub-teams.
This REU site is supported by the National Science Foundation and builds upon our previous REU site on robotic swarms. The current site is directed by Dr. Jason Gross (PI) and co-directed by Dr. Guilherme Pereira (Co-PI)
The site focuses on recruiting a diverse set of participants, mainly from the
Appalachian region, including students from underrepresented groups. All
students, from all regions of the U.S., interested in robotics are welcome to apply.
Projects
Enabling rural robotics and automation in the rural Appalachia will require new approaches which will be explored in several fundamental robotics research areas. The project areas will lead to innovation to address topics including: 1) robot programming by demonstration 2) estimation methods for under-canopy localization; 3) decision-making approaches for driving in forest trails; and 4) new legged robots for agriculture applications. During the selection process, REU applicants will be requested to list their top three interests from these research areas. More information about each area can be found below. We also encourage applicants to explore each mentor’s website to learn more about their individual research activities.
|
1) Robot programming by demonstration for agriculture applications
|
|
|
|
Recent research on robot programming by demonstration has allowed robots
to execute tasks by imitating the behavior of human beings. From system
identification to modern deep-neural networks, several techniques have
been used to program robots. This research topic will explore the use
of such technology for agricultural applications, where several generations
of knowledge have taught the human farmer how to perform their tasks
in effective and efficient ways. These tasks include produce picking
and processing, tree pruning, weed removal and animal milking. The students
will research and develop new hardware (grippers, teaching devices, etc)
and software (new learning methods and algorithms) that will allow a
commercial collaborative robot to be trained by a farmer on the solution
of important agricultural tasks.
Mentors: Dr. Guilherme Pereira and Dr. Dimas Dutra |
|
2) Reliable Legged Localization in Forests
|
|
|
|
Several applications would benefit from the use of robots in forested regions. This is complicated under forest canopies because the availability and quality of the Global Navigation Satellite System (GNSS) are hindered by the signal attenuation of dense forests. On the other hand, this presents an interesting problem set-up because GNSS is not completely unavailable for use, and it can be made available when going above tree cover. In this project students will explore estimation strategies for improving localization of robots operating in forested regions under canopy. In this project, students could have the flexibility to leverage advances in estimation algorithms alongside advances in robotic platforms. For example, students working with a legged robot could exploit zero-velocity updates inertial sensing alongside low quality GNSS signals to achieve a reliable solution.
Mentor: Dr. Jason Gross |
|
3) Autonomous Driving on Hiking Trail
|
|
|
|
The research on autonomous driving so far has been mainly focused on paved
roads and urban environments. Driving on common hiking trails offers
such challenges: complex geography, unmarked boundaries, GPS degradation,
interacting with pedestrians, bikers, and wild animals, changes of different
time scales (e.g., from falling trees to growing bushes), weather conditions,
among many others. Research in this direction, especially with physical
robot testing in complex environments, will advance the state of the
art in robot perception and decision-making methods. The REU research
will focus on semantic outdoor mapping and terrain traversability
assessment. The students will collect a large set of camera, LiDAR, GPS,
Inertial, and wheel encoder data when driving a robot on hiking trails.
They will perform real-time SLAM mapping and train image classifiers
to identify objects along the way, such as trails, trees, rocks, and
people.
Mentor: Dr. Yu Gu |
|
4) Legged Farm Equipment for Sloped Fields
|
|
|
|
To increase mobility in rural, unstructured environments, legged robots
are needed. Wheels perform best with infrastructure such as roads and
rails, which do not exist in some rural environments. In crop fields,
wheeled vehicles require wide gaps between rows of crops, which waste
soil that could be used for growing and becomes compacted by tractors
over time. Legged farm equipment would ameliorate these problems by replacing
wide ruts with small “stepping stones” among densely planted crops, standing
tall enough to have clearance over plants, and shifting their posture
and center of mass to operate safely on slopes. To realize this vision,
the energy efficiency and reliability of walking robots’ mechanics and
control must be improved. This project will prototype novel drivetrain
mechanisms with two key features: cable-actuation of distal leg joints;
and a spring-actuator combination. Prototypes are expected to be iterated
upon several times and subjected to through validation tests, including
but not limited to measurements and calculations of leg inertia, actuation
efficiency, and internal stresses. Control software must also be developed
in parallel.
Mentor: Dr. Nicholas Szczecinski |
Program Dates:
Summer 2024: May 20-July 26, 2024 (10 weeks in duration).
Summer 2025: May 19-July 25, 2025 (10 weeks in duration)
Participant Benefits:
Stipend of $7,000 ($700/week for 10-weeks), lodging, meal expenses, travel reimbursement to/from REU Site (limited to ~$500/participant) and comprehensive training to move participants toward intellectual and research independence.
Eligibility
- U.S. citizenship or permanent residency is required.
- Students must be currently enrolled in an undergraduate program (no specific majors are required).
- Students must not have completed an undergraduate degree prior to the summer program.
- Students from any higher-education institution of the U.S. are eligible, but students from institutions in the Appalachian region are especially encouraged to apply.