Technology Experiment: Landmark Based Navigation

Landmark based navigation is intended to greatly improve robot position estimation over dead reckoning by tracking visual features in the environment and using them as landmarks. This measurement returns bearing to the visual feature only. No a priori knowledge regarding the position of the landmarks is assumed, so the problem is one of simultaneously mapping and localizing. This is analogous to being placed in a foreign environment and using it to localize while, at the same time, building a model of that environment.

During this field season, we will use visual tracking software to process a sequence of panoramic images and store the bearing to each visual target in each frame over the sequence. In addition, we will process telemetry logs and extract steering and wheel encoder values. We will then use various estimation methods, such as Extended Iterated Kalman Filter, Covariance Intersection, and others, to localize the robot.

The current hardware configuration for landmark based navigation consists of a panoramic camera connected to a framegrabber on board a dual-Pentium PC mounted on Nomad, as well as wheel encoders and steering encoders on Nomad's chassis. The PC will be used to log images directly from the framegrabber, and to log NDDS messages containing telemetry while Nomad drives on the ice. Telemetry and images will be processed off-line to produce mapping/localization results.

 Back to Expedition Objectives

Robotic Search for Antarctic Meteorites 1998
All material on this page is property of NASA and Carnegie Mellon University. Any image or text
taken from this site and incorporated into another document without consent violates the Copyright
Law of the United States and the Berne International Copyright Agreement.
Send comments, questions, or suggestions to Dimitrios Apostolopoulos.
This document prepared by Michael Wagner.