eSSENCE@LU 4:5 - Bioinspired architectures for convolutional neural networks with applications in low light navigation and mapping

The main aim of the project is to develop artificial architectures based on existing primitive animal neural architectures. The workplan involves disciplinary and methodological research on the border between information science and biology. The goal is to investigate relevant biological models and modify these so that they become mathematically and algorithmically tractable to solve the underlying engineering problems robustly and accurately. The project specifically looks at how convolutional neural networks can be used in applications related to visual navigation, localization and recognition at low light levels.

The project acts as a bridge between biology, computer vision and mathematics.