Research Statement: My research is focused on developing collaborative agents for mixed-initiative systems that use both human and machine computation to solve problems quickly and reliably. I co-developed Legion, a system for real-time collaborative control of user interfaces. Legion employs a model that lets the crowd, a dynamic group of workers of varying reliability, act as a single reliable worker. I am currently working to develop human-machine hybrid versions of traditional AI solvers that are robust enough to be deployed in the real world. These systems are designed to work now because of crowd input but so that AI handle increasingly larger portions of the workload as it advances.

Agent Eyes Scribbler Turker Interface Robot Maze
Legion
Legion enables real-time crowd control of existing interfaces by making the crowd act as a single, more reliable, worker. While most current crowdsourcing systems enable offline human computation, Legion utilizes a continuous closed-loop interface with workers to allow real-time control. Each worker functions independantly, and Legion combines their input into a single control stream using a selectible input mediator.
Trans Ex LAR Word Prediction
Real-time Crowdsourcing for AT
Assistive techonology (AT) aims to help disabled users by enabling greater independance. However, tasks such as supplementing the users sensing capability is often closely tied to cognition, and as such, requires greater amounts of intelligence then currently available. We use the crowd to support greater functionality in these systems by leveraging human computation to fill in where automated systems and AI fall short.