Broertjes, M. and Jong, S. de (2009) Topological localization with omnidirectional images using an Echo State Network. Bachelor's Thesis, Artificial Intelligence.
|
Text
AI_Bc_2009_MBroertjes.pdf - Published Version Download (772kB) | Preview |
Abstract
We present a model for topological localization of robots equipped with hyperbolic mirrors for omnidirectional vision. Our goal is to achieve localization on a biologically plausible manner. The omnidirectional images are first pre-processed to data which can be used as input for an artificial neural network. The model uses an Echo State Network to perform a localization based on current input and past network activation. We test our model using a dataset of annotated omnidirectional images. We compare the results of our model with two baseline algorithms, namely a nearest neighbour algorithm and a Multi Layer Perceptron and with other research performed on the same dataset. We also show that a simple step in the pre-processing stage, which is inspired by SIFT and uses gradients, greatly improves the performance of the model.
| Item Type: | Thesis (Bachelor's Thesis) |
|---|---|
| Degree programme: | Artificial Intelligence |
| Thesis type: | Bachelor's Thesis |
| Language: | English |
| Date Deposited: | 15 Feb 2018 07:28 |
| Last Modified: | 15 Feb 2018 07:28 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/8572 |
Actions (login required)
 |
View Item |