Просмотр новости
Найдите то, что Вас интересует
Спутники
РН
SpaceX
Роскосмос
NASA
СпутникиРНSpaceXРоскосмосNASA
Highway construction zones produce transient pollutant plumes in corridors where static stations are not present to capture pollutant concentrations. Existing uncrewed aerial vehicle (UAV) solutions address mobility but treat each measurement as equally trustworthy, ignoring sensor saturation, drift, motion-induced airflow distortion, and time stamp offsets that systematically corrupt decisions where monitoring matters most. This paper proposes an integrity-aware UAV sensing framework that couples an onboard epistemic state with a cloud-based multi-agent integrity synthesis layer. Each UAV maintains a local measurement window characterized by adequacy metrics, i.e., outlier, stability, gradient, and integrity flags, saturation, limit-of-detection, drift, motion, synchronization, and adaptively extends its sampling window when any violation is detected, so that revisit-constrained measurements remain self-certified. Validated window summaries are transmitted over a long-range LoRa link and fused in the cloud to drive a cooperative multi-agent reinforcement learning (MARL) policy whose reward explicitly penalizes sampling actions tied to flagged windows. Field trials at two active Ohio Department of Transportation highway reconstruction sites in Akron, USA, demonstrate a 98% hotspot detection rate, 96% alignment with U.S. National Ambient Air Quality Standards (NAAQS) thresholds, and faster convergence than independent, single-agent, greedy, and random-walk baselines, with on-board decisions completing within 50 ms.
Mehdi Rahmati https://orcid.org/0000-0001-9012-3079
Article
8-17-2026
IEEE Journal on Miniaturization for Air and Space Systems (early access)
Highway construction zones produce transient pollutant plumes in corridors where static stations are not present to capture pollutant concentrations. Existing uncrewed aerial vehicle (UAV) solutions address mobility but treat each measurement as equally trustworthy, ignoring sensor saturation, drift, motion-induced airflow distortion, and time stamp offsets that systematically corrupt decisions where monitoring matters most. This paper proposes an integrity-aware UAV sensing framework that couples an onboard epistemic state with a cloud-based multi-agent integrity synthesis layer. Each UAV maintains a local measurement window characterized by adequacy metrics, i.e., outlier, stability, gradient, and integrity flags, saturation, limit-of-detection, drift, motion, synchronization, and adaptively extends its sampling window when any violation is detected, so that revisit-constrained measurements remain self-certified. Validated window summaries are transmitted over a long-range LoRa link and fused in the cloud to drive a cooperative multi-agent reinforcement learning (MARL) policy whose reward explicitly penalizes sampling actions tied to flagged windows. Field trials at two active Ohio Department of Transportation highway reconstruction sites in Akron, USA, demonstrate a 98% hotspot detection rate, 96% alignment with U.S. National Ambient Air Quality Standards (NAAQS) thresholds, and faster convergence than independent, single-agent, greedy, and random-walk baselines, with on-board decisions completing within 50 ms.
Hazarika, Ananya; Mahadik, Gaurav; Asare, Stephen; Masanja, Ntemi; Rahmati, Mehdi; Owusu-Danquah, Josiah; and Kidando, Emmanuel, "UAV-Based Intelligent Environmental Monitoring of Pollutants in Highway Construction Zones" (2026). Electrical and Computer Engineering Faculty Publications. 544.
https://engagedscholarship.csuohio.edu/enece_facpub/544
10.1109/JMASS.2026.3724380
DOWNLOADS
Since August 20, 2026
COinS