Currently, municipalities assess rolling noise on road surfaces using Close-Proximity measurements (CPX). To avoid these labor-intensive measurements, an opportunistic approach based on commodity sensors in a fleet of cars, is proposed. Blind sensor calibration eliminates the effect of measurement vehicle and varying observation conditions.Calibration relies on spatial coherence: modifiers and confounders do not interact strongly with location while the quantity of interest depends on location and less on measurement vehicle. Generalized additive speed models, car offset and de-noising autoencoders (DAE) were investigated. DAE achieves prominent results: (1) ratio of variability of measurements at a single location to the variability of measurements over all locations increases, (2) convergence of mean measurement at a location is faster, and (3) seasonal effects are eliminated. Finally, although the proposed method includes a diversity of tires, below 1600 Hz its results differ from CPX less than the difference between bi-annually repeated CPX measurements.
In real life, acoustic scenes and audio events are naturally correlated. Humans instinctively rely on fine-grained audio events as well as the overall sound characteristics to distinguish diverse acoustic scenes. Yet, most previous approaches treat acoustic scene classification (ASC) and audio event classification (AEC) as two independent tasks. A few studies on scene and event joint classification either use synthetic audio datasets that hardly match the real world, or simply use the multi-task framework to perform two tasks at the same time. Neither of these two ways makes full use of the implicit and inherent relation between fine-grained events and coarse-grained scenes. To this end, this paper proposes a relation-guided ASC (RGASC) model to further exploit and coordinate the scene-event relation for the mutual benefit of scene and event recognition. The TUT Urban Acoustic Scenes 2018 dataset (TUT2018) is annotated with pseudo labels of events by a simple and efficient audiorelated pre-trained model PANN, which is one of the state-ofthe-art AEC models. Then, a prior scene-event relation matrix is defined as the average probability of the presence of each event type in each scene class. Finally, the two-tower RGASC model is jointly trained on the real-life dataset TUT2018 for both scene and event classification. The following results are achieved. 1) RGASC effectively coordinates the true information of coarsegrained scenes and the pseudo information of fine-grained events.2) The event embeddings learned from pseudo labels under the guidance of prior scene-event relations help reduce the confusion between similar acoustic scenes. 3) Compared with other (nonensemble) methods, RGASC improves the scene classification accuracy on the real-life dataset.
Assessing road degradation typically requires specialized hardware (such as laser profilometers) or labor-intensive visual inspection. To facilitate large-scale, timely inspection of road surfaces, opportunistic sensing is proposed: Sound and vibration measurements are obtained from vehicles that are on the road for other purposes than measuring road quality. Prior work has addressed the problem of calibration and measurement noise removal from this abundance of measurements for a small number of measurement vehicles that drive on the same roads. However, as the deployment of opportunistic monitoring progresses, the applied techniques suffer from scalability. Here, a scalable self-supervised calibration and confounder removal (SCCR) algorithm is introduced. It allows to self-calibrate even if the data collection is done in distinct geographic areas and is capable of generalizing to vehicles not encountered during the training phase.Several model design alternatives are explored. After the application of SCCR, supervised training on a small subset of roads allows to predict observations made by standardized techniques also in areas where the latter have not been performed. The approach is tested and validated with 41 cars driving on 23,000 km of roads.
This publication presents the proceedings of TRA2020, the 8 th Transport Research Arena, which was planned to be held on 27-30 April 2020 in Helsinki. The physical conference event was cancelled due to the COVID-19 pandemic.
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.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.