Robot navigation using ultrasonic feedback

This thesis presents an autonomous navigation program for West Virginia University's mobile robot platform ANT. The program allows the ANT to find a predefined goal by comparing sensor data with a given map of the environment. ANT's ultrasonic sensors were used to sample the surroundings continuously, and this information was combined with dead reckoning data to build a two-dimensional map of the environment. Borenstein's Histogramic In-Motion Mapping technique was used to build sensor maps, which were analyzed to find features such as walls and corners. The features were matched with features in the given map to correct dead reckoning errors. Partial success was demonstrated in this work. However it was concluded that more than seven sensors are necessary when operating in an environment with smooth walls. The failure modes for the robot were analyzed and suggestions are made to overcome some of the sensing problems.

Name thesis:degree_name

MS

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Mechanical and Aerospace Engineering

Year dc:date.available

1999

Author dc:creator

• Baba, Akihiko

Contributors dc:contributor

• Larry E. Banta.

• Mechanical engineering
• Computer science

Identifier

https://researchrepository.wvu.edu/etd/942

OAI identifier oai:identifier

oai:researchrepository.wvu.edu:etd-1945