In this work, novelty detection identifies salient image features to guide autonomous robotic exploration. There is little advance knowledge of the features in the scene or the proportion that should count as outliers. A new algorithm addresses this ambiguity by modeling novel data in advance and characterizing regular data at run time. Detection thresholds adapt dynamically to reduce misclassification risk while accommodating homogeneous and heterogeneous scenes. Experiments demonstrate the technique on a representative set of navigation images from the Mars Exploration Rover "Opportunity." An efficient image analysis procedure filters each image using the integral transform. Pixel-level features are aggregated into covariance descriptors that represent larger regions. Finally, a distance metric derived from generalized eigenvalues permits novelty detection with kernel density estimation. Results suggest that exploiting training examples of novel data can improve performance in this domain.
展开▼
