Nathan Powell, RPI Graduate Student

 

Nathan Powell, RPI Graduate Student

Carnegie 113

February 6, 2019 12:00 PM - 1:30 PM

Humans and other animals are well adapted to actively seek out relevant visual information when performing tasks that require them to navigate complex environments. Whether it be a human avoiding another pedestrian on a busy street, or a dragonfly maintaining view of its prey, actively sensing the environment through vision allows for quick, efficient, and robust perceptual capabilities and motor responses. Recently, there has been an increase in the number of studies inspired by biological vision to create control algorithms for navigation in artificial and robotic systems. However, the computational power for real-time implementation of these models is limited. Here I review the potential importance of uncertainty reduction through active gaze (e.g. eye-movements) for real-time sensorimotor control in artificial systems, with a focus on small aerial robots. I then lay the groundwork for implementing uncertainty reduction through active gaze within an existing neurally-inspired algorithm for estimating heading direction and object motion based solely on visual input.

 

 

Add to calendar
Share|