It is shown that, in a limited domain, ultrasonic rangers can be used with model-based reasoning to identify one or two objects represented in the return signal. It is also shown that characteristics that are useful in helping discriminate include average amplitude times distance, standard deviation of the amplitude as a percentage of the maximum amplitude, and characteristics of the echo signal such as maximum echo amplitude times the square of the distance. It was found many of the characteristics (such as maximum amplitude) were not helpful in discriminating between objects at various distances. It was found that a list of line features describing perimeter shape was not useful when objects were considered at various distances. The results obtained are pertinent to the Autonomous Land Vehicle (ALV).<<ETX>>
[1]
Theodore P. Baker,et al.
Special Feature: An Architecture for Real-Time Software Systems
,
1986,
IEEE Software.
[2]
Hans P. Moravec,et al.
High resolution maps from wide angle sonar
,
1985,
Proceedings. 1985 IEEE International Conference on Robotics and Automation.
[3]
Anita M. Flynn,et al.
Redundant Sensors for Mobile Robot Navigation
,
1985
.
[4]
B Everett.
A Multi-Element Ultrasonic Ranging Array
,
1985
.
[5]
Sargur N. Srihari,et al.
A strategy for diagnosis based on empirical and model knowledge
,
1987
.