In this thesis sparse models are used to develop efficient solutions for the detection of objects and the navigation in an unknown environment.
Depth-Adaptive Superpixels are a novel image segmentation technique for combined colour and depth sensors which creates excellent oversegmentations and full segmentations for static images or video streams.
Event-based SLAM is a novel event-based simultaneous localization and mapping algorithm which uses a biologically inspired dynamic vision sensor to dramatically decrease the computational complexity of the problem.
«