Research Interests:
Robotics and AI, especially multi-robot systems, human-robot interaction, machine learning, multi-agent systems, and educational robotics.
Research Description:
My research vision is to build a complex human-robot system, in which humans and robots can cooperate seamlessly in the real world, with robot assisting humans to accomplish tasks, learning from experience, and adapting to new situations. My current research aims towards building multi-robot coalitions to accomplish complex applications. With the assistance from human operators, the overall human-robot team can achieve the robustness, effectiveness, and flexibility of the team solution in uncertain and dynamic situations.
My dissertation research focused on software reconfigurability on mobile robots. The problem is: given a heterogeneous robot team and a task, how to generate task solutions that make the most efficient use of the robot team. To attack this challenge, we have designed and developed ASyMTRe (pronounced "asymmetry"), which stands for Automated Synthesis of Multi-Robot Task Solutions through Software Reconfiguration. We have proved that ASyMTRe can be applied to a lot of multi-robot applications, such as multi-robot transportation and box-pushing tasks, and our future work would involve applying ASyMTRe to human-robot teams.
Previously, I was involved in a project called SDR (Software for Distributed Robotics) which successfully demonstrated the deployment a large number of resource-bounded robots (70+) with only a few capable robots (with more sensing capabilities), into a distributed sensor network to perform reconnaissance and surveillance. This DARPA project was a joint effort between Science Applications International Corporation (SAIC), The University of Tennessee, and the University of Southern California.
My responsibilities included:
- Using CMUcam to implement leader-follower formation for navigation.
- Designed a fiducial marker mounted on resource-bounded robots, so that more capable Pioneer robots with a camera can recognize the robot by determining the marker's id, orientation, angle, and distance.
Current Research Topics:
- Time-extended assignment: planning strategies are needed for robots to schedule a list of tasks.
- Peer-to-peer human-robot collaboration in a search and rescue scenario.
- Genetic algorithms for automating the multi-robot team solution process.
Potential Research Topics:
- Task Allocations:
- Combinatorial auctions of tasks: multiple tasks are allocated and each participant can bid on any combination of the tasks.
- Constrained-based task allocation: tasks are constrained with one another. For example, one task needs to be completed before another task can start or two tasks must be started simultaneously.
- Task decomposition: when a complex mission is assigned, how to decompose it to primitive or simple tasks? Decompose-then-allocate method or allocate-then-decompose method?
- Allocation of tasks that require tight coordination or complex constraints.
- Reallocation of tasks and replanning in uncertain environment.
- Human-Robot Cooperation:
- Modeling of human capabilities to facilitate the interaction between humans and robots.
- Building the communication channel between humans and robots.
- Machine Learning:
- Adapting the utility function (which is used to characterize the solution quality of the task allocation problem) to the changing environment and team composition.
- Computer Vision:
- Gesture recognition for human-robot interaction.
- Robot recognition using artificial markers.
- Others:
- Simultaneous localization and mapping (SLAM).
- Topological map building.
- Swarm intelligence: simple agents interact locally with each other that lead to the emergence of global behavior, such as ant colonies.
I graduated from UTK DILab in August 2006 and my advisor is Dr. Lynne E. Parker.
