- Abstract:
-
Doppler-shifts are a rich source of information not exploited by commercial ultrasonic range sensors for robots. Bats use Doppler-shifts for several tasks such as obstacle avoidance and prey capture, hence our motivation for incorporating the use of Doppler to mobile robot navigation. In this work we compare two different cochlear filterbank models (linear and gamma-tone) together with two methods (artificial neural network and weighted average) for extracting the Doppler information included in the echoes received by RoBat, a biomimetic sonarhead mounted on a mobile robot, when navigating. Experiments with two different type of reflectors --- a post and a wall --- for each of the methods under test are performed. Some of the configurations gave overall good performance, and others showed an interesting and incorrect behaviour which was identified after further investigations as a dynamic range problem in the sonarhead.
- Copyright:
- 2002 by The University of Edinburgh. All Rights Reserved
- Links To Paper
- No links available
- Bibtex format
- @InProceedings{EDI-INF-RR-0058,
- author = {
Jose Carmena
and John Hallam
},
- title = {A comparison of methods for estimating Doppler-shift using bat-inspired cochlear filterbank models},
- book title = {Fourth European workshop on advanced mobile robots (EUROBOT'01)},
- year = 2001,
- month = {Sep},
- pages = {9-16},
- }
|