More Like this

1. Road Condition Monitoring Using Vehicle Built-in Cameras and GPS Sensors: A Deep Learning Approach

   https://doi.org/10.3390/vehicles5030051  

   Ruseruka, Cuthbert; Mwakalonge, Judith; Comert, Gurcan; Siuhi, Saidi; Perkins, Judy (September 2023, Vehicles)

   Road authorities worldwide can leverage the advances in vehicle technology by continuously monitoring their roads’ conditions to minimize road maintenance costs. The existing methods for carrying out road condition surveys involve manual observations using standard survey forms, performed by qualified personnel. These methods are expensive, time-consuming, infrequent, and can hardly provide real-time information. Some automated approaches also exist but are very expensive since they require special vehicles equipped with computing devices and sensors for data collection and processing. This research aims to leverage the advances in vehicle technology in providing a cheap and real-time approach to carry out road condition monitoring (RCM). This study developed a deep learning model using the You Only Look Once, Version 5 (YOLOv5) algorithm that was trained to capture and categorize flexible pavement distresses (FPD) and reached 95% precision, 93.4% recall, and 97.2% mean Average Precision. Using vehicle built-in cameras and GPS sensors, these distresses were detected, images were captured, and locations were recorded. This was validated on campus roads and parking lots using a car featured with a built-in camera and GPS. The vehicles’ built-in technologies provided a more cost-effective and efficient road condition monitoring approach that could also provide real-time road conditions. 

   more » « less
2. Eye in the Sky: 360° Inspection of Bridge Infrastructure Using Uncrewed Aerial Vehicles (UAVs)

   https://doi.org/10.1177/03611981231185138  

   Congress, Surya_Sarat Chandra; Escamilla, Jesse; Chimauriya, Hiramani; Puppala, Anand J (April 2024, Transportation Research Record: Journal of the Transportation Research Board)

   Bridges play a key role in supporting the transportation network in the United States. The 2021 infrastructure report card prepared by ASCE highlighted that more than 40% of bridges in the U.S. are over 50 years old. Some of these bridges are classified as structurally deficient, even though they are safe to travel. To address these challenges, highway agencies are exploring innovative technologies to conduct inspections and realize benefits in relation to access, cost, and safety. Federal and state DOTs have conducted several studies on the application of uncrewed aerial vehicles (UAVs) for bridge health monitoring. This study identified the existing knowledge gap in performing 360° inspection of bridges. In this current research, UAVs were demonstrated for conducting 360° inspections of three different bridges in Alaska. The locations of the aerial images during the inspections were also pictographically represented to provide a holistic idea for the highway agencies and practitioners. Three-dimensional models representing the actual conditions of the bridge were generated and used for comparing the bridge condition assessments with traditional inspection reports. Infrared imagery was also collected to identify the effect of thermal loading in assessing the conditions of the bridge elements. The applicability and recommendation scale for the use of UAVs for different bridge inspections was provided. The approach demonstrated in this study is expected to result in more than 90% savings in storage requirements and contribute to an increase in the applications of UAVs for conducting 360° bridge inspections across the U.S. 

   more » « less
3. DeQA: On-Device Question Answering

   Cao, Qingqing; Weber, Noah; Balasubramanian, Niranjan; Balasubramanian, Aruna (June 2019, ACM International Conference on Mobile Systems, Applications, and Services)

   Today there is no effective support for device-wide question answer- ing on mobile devices. State-of-the-art QA models are deep learning behemoths designed for the cloud which run extremely slow and require more memory than available on phones. We present DeQA, a suite of latency- and memory- optimizations that adapts existing QA systems to run completely locally on mobile phones. Specifi- cally, we design two latency optimizations that (1) stops processing documents if further processing cannot improve answer quality, and (2) identifies computation that does not depend on the ques- tion and moves it offline. These optimizations do not depend on the QA model internals and can be applied to several existing QA models. DeQA also implements a set of memory optimizations by (i) loading partial indexes in memory, (ii) working with smaller units of data, and (iii) replacing in-memory lookups with a key-value database. We use DeQA to port three state-of-the-art QA systems to the mobile device and evaluate over three datasets. The first is a large scale SQuAD dataset defined over Wikipedia collection. We also create two on-device QA datasets, one over a publicly available email data collection and the other using a cross-app data collection we obtain from two users. Our evaluations show that DeQA can run QA models with only a few hundred MBs of memory and provides at least 13x speedup on average on the mobile phone across all three datasets. 

   more » « less
4. Application of Satellite-Based Remote Sensing for the Management of Pavement Infrastructure Assets

   https://doi.org/10.1177/03611981241230535  

   Gajurel, Amit; Puppala, Anand J; Biswas, Nripojyoti; Chimauriya, Hiramani Raj (September 2024, Transportation Research Record: Journal of the Transportation Research Board)

   The long-term monitoring of transportation infrastructure assets at a lower cost and with short mobilization time is of significant interest to both state and federal transportation agencies in the U.S. Because of the significant improvement in spatial and temporal resolution of synthetic aperture radar (SAR) remote sensing systems and a notable reduction in the cost of data acquisition, SAR has now become a viable method to provide economic and rapid condition assessment of transportation assets. A research study was developed and performed to comprehensively perform the inspection and characterization of a pavement surface based on the amplitude of backscattering of an X-band radar. In situ characterization of the test site was first performed using traditional inertial profilers and aerial photogrammetry with unmanned aerial vehicle (UAV) surveys. The results from these in situ methods were compared with the corrected amplitude of the SAR data, which indicated that the distribution of surface roughness values computed from the inertial profiler, UAV, and SAR exhibited similar probability densities at various segmental lengths considered in this study. This suggested that the problematic areas that are evident during in situ characterization can be delineated and quantified based on the normalized radar cross section of the pavement surface. Overall, the outcome of this research exhibits the potential of SAR for future transportation asset management undertakings, and the systematic framework developed as a part of this research could be of significant interest to engineers and transportation practitioners. 

   more » « less
5. Improving Label Noise Robustness with Data Augmentation and Semi-Supervised Learning (Student Abstract)

   https://doi.org/10.1609/aaai.v35i18.17924  

   Nishi, Kento; Ding, Yi; Rich, Alex; Höllerer, Tobias (May 2021, Proceedings of the AAAI Conference on Artificial Intelligence)

   Modern machine learning algorithms typically require large amounts of labeled training data to fit a reliable model. To minimize the cost of data collection, researchers often employ techniques such as crowdsourcing and web scraping. However, web data and human annotations are known to exhibit high margins of error, resulting in sizable amounts of incorrect labels. Poorly labeled training data can cause models to overfit to the noise distribution, crippling performance in real-world applications. In this work, we investigate the viability of using data augmentation in conjunction with semi-supervised learning to improve the label noise robustness of image classification models. We conduct several experiments using noisy variants of the CIFAR-10 image classification dataset to benchmark our method against existing algorithms. Experimental results show that our augmentative SSL approach improves upon the state-of-the-art. 

   more » « less