PeRL STUDIES AUTONOMOUS NAVIGATION & MAPPING FOR MOBILE ROBOTS IN A PRIORI UNKNOWN ENVIRONMENTS.

At a Glance

Synopsis

PeRL studies autonomous navigation and mapping for mobile robots in a priori unknown environments.

Dig Deeper

Our work

The Perceptual Robotics Laboratory (PeRL) at the University of Michigan studies problems related to autonomous navigation and mapping for mobile robots in a priori unknown environments.


Visually Augmented Navigation (VAN) derived 3D model of the RMS Titanic using monocular imagery collected from a ROV.

The goal of this work is to enable robots with the ability to autonomously navigate and map their environment, recognizing previously visited places much as a human would.

Since GPS does not work underwater, underground, on other planets, or even inside buildings, solving this problem is critical to developing practical, capable, autonomous mobile robots.


The University of Michigan's two 100m-rated Ocean-Server AUV's provide a platform for real-time VAN algorithm testing and validation.

To study this problem, the research methodology within PeRL balances theory with experimental validation developing algorithms (software) in the areas of underwater computer vision and image processing, Bayesian filtering and smoothing, and systems engineering, in conjunction with new platform development (hardware) such as time-synchronized acoustic navigation systems, Autonomous Underwater Vehicles (AUVs), and ground robotics.

Current PeRL projects include developing an automated Ford Fusion, autonomous ship-hull inspection for the Navy, multi-AUV cooperative navigation, active safety situational awareness for automotive vehicles, large-area acoustic and optical simultaneous localization and mapping (SLAM), and the design of a multi-AUV SLAM testbed. Most recently we have begun developing a dual ground/aerial robotics testbed as part of the Naval Engineering Education Center.