Incremental learning for Adaptive Visual Place Recognition in Dynamic Indoor Environments
Jie Luo, Barbara Caputo, Patric Jensfelt, Andrzej Pronobis · 2006
Vision-based place recognition is a desirable feature for an autonomous mobile system. In order to work in realistic scenarios, a visual recognition algorithm should have two key properties: robustness and adaptability. This thesis focuses on the latter, and presents a discriminative incremental learning approach to place recognition. We propose a solution based on incremental extensions of support vector machine classifier. Computational and memory efficiency are crucial for mobile robot platforms aiming to continuously work in real-world settings; thus we put the emphasis on these properties. We use a recently introduced memory-controlled incremental technique, which allows to control the memory requirements as the system updates its internal representation. At the same time, it preserves the recognition performance of the batch algorithm and runs online. In order to assess the method, we acquired a database capturing the intrinsic variability of places over time. Extensive experiments show the power and the potential of the approach.