Using Deep Autoencoders to Investigate Image Matching in Visual Navigation
This paper discusses the use of deep auto encoder networks to find a compressed representation of an image, which can be used for visual navigation. Images reconstructed from the compressed representation are tested to see if they retain enough information to be used as a visual compass (in which an image is matched with another to recall a bearing/movement direction) as this ability is at the heart of a visual route navigation algorithm. We show that both reconstructed images and compressed representations from different layers of the auto encoder can be used in this way, suggesting that a compact image code is sufficient for visual navigation and that deep networks hold promise for finding optimal visual encodings for this task.
KeywordsVisual navigation Insect-inspired robotics Deep neural network Autoencoder
This work was funded by the Newton Agri-Tech Program RICE PADDY project (no. STDA00732). AP and PG were also funded by EPSRC grant EP/P006094/1.
- 1.Graham, P., Philippides, A.: Insect-inspired vision and visually guided behavior. In: Bhushan, B., Winbigler, H.D. (eds.) Encyclopedia of Nanotechnology. Springer, Netherland (2015)Google Scholar
- 17.Bengio, Y., Courville, A.C., Vincent, P.: Unsupervised feature learning and deep learning: A review and new perspectives. CoRR, abs/1206.5538 (2012)Google Scholar
- 18.Kingma, D., Ba, J.: Adam: A method for stochastic optimization. arXiv:1412.6980 (2014)
- 19.Philippides, A., Graham, P., Baddeley, B., Husbands, P.: Using neural networks to understand the information that guides behavior: a case study in visual navigation. Artif. Neural Netw. 1260, 227–244 (2015)Google Scholar
- 21.Baddeley, B., Graham, P., Philippides, A., Husbands, P.: Models of visually guided routes in ants: embodiment simplifies route acquisition. In: Jeschke, S., Liu, H., Schilberg, D. (eds.) ICIRA 2011. LNCS, vol. 7102, pp. 75–84. Springer, Heidelberg (2011). doi: 10.1007/978-3-642-25489-5_8 CrossRefGoogle Scholar