Gabel Family Term Assistant Professor
When multiple robots coordinate to do complex tasks, the task allocation, or the assignment of tasks to robots, has a large effect on the system performance. We consider distributed task allocation for a team of robots. Specifically, we analyze task switching as a method for improving a task allocation as the system is running. For situations where computing even a locally optimal task allocation would be too expensive, we have designed heuristics that nonetheless guarantee task completion. We additionally consider how historical information about such a system's performance could be used to improve future allocations. We have proposed an algorithm for partitioning the environment into regions of equal workload for the robots and used hubs to help robots pass tasks to each other. We have tested our algorithms both in simulation and in hardware experiments.
Joint work with Nora Ayanian and Daniela RusiDiary: Semantic Compression for Big Data
We developed iDiary, a system that turns large GPS signals collected from smartphones into textual descriptions of the trajectories. We have designed algorithms for semantic compression in order to compute the locations commonly visited by a user. Specifically, we have defined the (k,m)-segment mean trajectory clustering problem and built coresets to allow fast computation of a (1+ε)-approximation to this problem. iDiary uses an external database to map the identified locations to textual descriptions and activities. It features a user interface similar to Google Search that allows users to type text queries on their activities (e.g., "Where did I buy books?") and receive textual answers based on their GPS data.
Joint work with Dan Feldman, Andrew Sugaya, and Daniela Rus