Research

In Robotic Search Lab (RSL), we explore search problems from both biological and robotic perspectives. The former one is called human search and the latter one is called robot search.

Human Search:

The Human brain is an amazing machine. It can solve problems under complex environments. Search is one of daily human activities. The goal of human search is to explore how humans search for targets so efficiently. Human subjects search for the target via teleoperation. Human data (e.g., gaze, control, robot position and steaming images from the robot) is stored for analysis. We analyze humans' decision-making processes; on the other hands, robots can imitate humans' search via learning.

Robot Search:

Autonomous search is one of NP-hard problems in operation research and computer science. Scientists tried to find an optimal solution since world war II. But, no one succeeded. The goal of robotic search is to enable robots to search for targets as soon as possible. It includes coverage problems, probabilistic search and minimum-time trajectory planning. All of them are NP-hard problems. How to simultaneously solve these problems within a theoretical bound is a challenge. We utilize submodularity of search problems to enable robots to search for targets efficiently.

RSL members and Yi-Ta Ho (UAV pilot)

We hired UAV pilots to do human experiments each summer and tried to learn how their brains work on search problems.

Google Sites

Report abuse