In this paper, we study a problem encountered during our ongoing efforts to locate radio-tagged fish aggregations with robots. The problem lies at the intersection of search-based methods whose objective is to detect a target, and active target localization methods whose objective is to precisely localize a target given its initial estimate. Real-world sensing constraints such as limited and unknown range, large measurement time, and ambiguity in bearing measurements make it imperative to have an intermediate initialization phase to transition from search to localization. We present a local search strategy aimed at reliably initializing an estimate for a single target based on observations from field experiments. We then extend this strategy to initialize multiple targets, exploiting the proximity of nearby aggregated tagged fish to decrease the cost of initialization per target. We evaluate the performance of our algorithm through simulations and demonstrate its utility through a field experiment where the robot successfully detects, initializes and then localizes nearby targets.
Although state-of-the-art object detection methods have shown compelling performance, models often are not robust to adversarial attacks and out-of-distribution data. We introduce a new dataset, Natural Adversarial Objects (NAO), to evaluate the robustness of object detection models. NAO contains 7,934 images and 9,943 objects that are unmodified and representative of real-world scenarios, but cause state-of-the-art detection models to misclassify with high confidence. The mean average precision (mAP) of EfficientDet-D7 drops 74.5% when evaluated on NAO compared to the standard MSCOCO validation set.Moreover, by comparing a variety of object detection architectures, we find that better performance on MSCOCO validation set does not necessarily translate to better performance on NAO, suggesting that robustness cannot be simply achieved by training a more accurate model.We further investigate why examples in NAO are difficult to detect and classify. Experiments of shuffling image patches reveal that models are overly sensitive to local texture. Additionally, using integrated gradients and background replacement, we find that the detection model is reliant on pixel information within the bounding box, and insensitive to the background context when predicting class labels. NAO can be downloaded here.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.