Search for: All records

The deep neural network (DNN) model for computer vision tasks (object detection and classification) is widely used in autonomous vehicles, such as driverless cars and unmanned aerial vehicles. However, DNN models are shown to be vulnerable to adversarial image perturbations. The generation of adversarial examples against inferences of DNNs has been actively studied recently. The generation typically relies on optimizations taking an entire image frame as the decision variable. Hence, given a new image, the computationally expensive optimization needs to start over as there is no learning between the independent optimizations. Very few approaches have been developed for attacking online image streams while taking into account the underlying physical dynamics of autonomous vehicles, their mission, and the environment. The article presents a multi-level reinforcement learning framework that can effectively generate adversarial perturbations to misguide autonomous vehicles’ missions. In the existing image attack methods against autonomous vehicles, optimization steps are repeated for every image frame. This framework removes the need for fully converged optimization at every frame. Using multi-level reinforcement learning, we integrate a state estimator and a generative adversarial network that generates the adversarial perturbations. Due to the reinforcement learning agent consisting of state estimator, actor, and critic that only uses image streams, the proposed framework can misguide the vehicle to increase the adversary’s reward without knowing the states of the vehicle and the environment. Simulation studies and a robot demonstration are provided to validate the proposed framework’s performance. 

Catastrophic release of heavy metals from the King River mine in Colorado and the Minas Gerais dam in Brazil have brought to the forefront the importance of contaminant stabilization and remediation in surface waters. Permeable reactive materials are currently utilized for the remediation of heavy metals and other pollutants by employing reactive media to remove contaminants. This research investigated the use of fly ashes with loss on ignition or sulfur trioxide exceeding ASTM C618 limits to enhance pollutant removal in pervious concrete. The high carbon and sulfur contents of the noncompliant fly ashes provide additional capacity to remove lead, cadmium, and zinc. High-sulfur and high-carbon fly ashes were less effective in metal removal at higher metal concentrations but improved removal at lower concentrations. These results suggest pervious concrete can be designed as an effective remedial technique for use in many infrastructure applications, including beneath permeable pavement, permeable asphalt, revetment, permeable shoulders, gabions for slope stability, mine tailing dams, and emergency surface water cleanup. 

Permeable reactive barriers (PRBs) are a well-known technique for groundwater remediation using industrialized reactive media such as zero-valent iron and activated carbon. Permeable reactive concrete (PRC) is an alternative reactive medium composed of relatively inexpensive materials such as cement and aggregate. A variety of multimodal, simultaneous processes drive remediation of metals from contaminated groundwater within PRC systems due to the complex heterogeneous matrix formed during cement hydration. This research investigated the influence coarse aggregate, portland cement, fly ash, and various combinations had on the removal of lead, cadmium, and zinc in solution. Absorption, adsorption, precipitation, co-precipitation, and internal diffusion of the metals are common mechanisms of removal in the hydrated cement matrix and independent of the aggregate. Local aggregates can be used as the permeable structure also possessing high metal removal capabilities, however calcareous sources of aggregate are preferred due to improved removal with low leachability. Individual adsorption isotherms were linear or curvilinear up, indicating a preferred removal process. For PRC samples, metal saturation was not reached over the range of concentrations tested. Results were then used to compare removal against activated carbon and aggregate-based PRBs by estimating material costs for the remediation of an example heavy metal contaminated Superfund site located in the Midwestern United States, Joplin, Missouri. 

Heavy metals contaminants include lead, chromium, arsenic, zinc, cadmium, copper, and mercury all of which can cause significant damage to human health and the environment as a result of their mobility and solubility within groundwater. In the Midwest portion of the United States, soil and groundwater based lead, zinc, and cadmium are the prominent pollutants of concern. While remedial technologies exist for heavy metals pollution, the majority of these solutions are expensive to design, maintain, and install. Drinking water treatment waste (DWTW) is a currently landfilled, relatively pure, industrial waste byproduct composed almost entirely of calcium oxide produced during water purification processes. Measured doses of drinking water treatment waste were submerged in synthetic groundwater solutions containing 0.01, 0.1, and 1.0 millimolar concentrations of lead, cadmium and zinc in order to determine if this material could provide remedial measures for heavy metals. In addition, the geomechanical properties and chemical composition of the material were determined. Removal rates varied based upon internal and external water content as well as flocculant formation. However, all tests verify that the material is capable of heavy metals removal at relatively rapid rates. This data suggests that when entrained in a previous matrix, the reactive nature of the byproduct sorbs ions in solution passing through the matrix.