Towards semi-supervised learning of semantic spatial concepts

Abstract

The ability of building robust semantic space representations of environments is crucial for the development of truly autonomous robots. This task, inherently connected with cognition, is traditionally achieved by training the robot with a supervised learning phase. We argue that the design of robust and autonomous systems would greatly benefit from adopting a semi-supervised online learning approach. Indeed,the support of open-ended, lifelong learning is fundamental in order to cope with the dazzling variability of the real world, and online learning provides precisely this kind of ability. Here we focus on the robot place recognition problem, and we present an online place classification algorithm that is able to detect gap in its own knowledge based on a confidence measure. For every incoming new image frame, the method is able to decide if (a) it is a known room with a familiar appearance, (b) it is a known room with a challenging appearance, or (c) it is a new, unknown room. Experiments on a subset of the challenging COLD database show the promise of our approach.

Reliable topological place detection in bubble space

“…There has been numerous work that have addressed this issue. One approach is to detect challenging instances in the recognition module where the detection is based on confidence measures associated with label assignment process [13]. Another approach is to use statistical approaches to changepoint detection [14], [15].…”

Towards generalization of semi-supervised place classification over generalized Voronoi graph

“…For example, Martinez-Gomez and Caputo suggested a subdivision of rooms in terms of their appearance, the activities people usually perform in them, and the objects they contain [33]. These systems usually require rich sensory modalities and hierarchical concept modeling, so that a robot can integrate its understanding about distinct topological areas with its knowledge about the presence of certain objects [17].…”

“…For this purpose, a few variations of the original SVM algorithm have been proposed. For example, algorithms like online independent-SVM (OISVM) and memory-controlled incremental SVM do not require storing all incoming data, and have selection mechanisms to guarantee a bounded memory growth [11,26,33]. These approaches focus more on the algorithmic efficiency and can be further improved by considering the spatial context.…”