100 research outputs found
Detecting Attacks Against Deep Reinforcement Learning for Autonomous Driving
No full textWith the advent of deep reinforcement learning, we witness the spread of novel autonomous driving agents that learn how to drive safely among humans. However, skilled attackers might steer the decision-making process of these agents through minimal perturbations applied to the readings of their hardware sensors. These force the behavior of the victim agent to change unexpectedly, increasing the likelihood of crashes by inhibiting its braking capability or coercing it into constantly changing lanes. To counter these phenomena, we propose a detector that can be mounted on autonomous driving cars to spot the presence of ongoing attacks. The detector first profiles the agent's behavior without attacks by looking at the representation learned during training. Once deployed, the detector discards all the decisions that deviate from the regular driving pattern. We empirically highlight the detection capabilities of our work by testing it against unseen attacks deployed with increasing strength
Warming exerts greater impacts on subsoil than topsoil CO2 efflux in a subtropical forest
No full textCorrection: Pore-size-tuned host–guest interactions in Co-MOFs via in situ microcalorimetry: adsorption and magnetism
No full textCorrection for ‘Pore-size-tuned host–guest interactions in Co-MOFs via in situ microcalorimetry: adsorption and magnetism’ by Chengfang Qiao et al., J. Mater. Chem. C, 2017, 5, 1064–1073.</p
Covert-inspired flow control devices for lift enhancement and stall mitigation
No full textUnmanned aerial vehicles (UAVs) are taking on more expansive roles in military and commercial applications. However, their adaptability and agility are still inferior to that of their biological counterparts like birds, especially at low and moderate Reynolds numbers. A system of aeroelastic devices used by birds, known as the covert feathers, has been considered as a natural flow-control device for mitigating flow separation, enhancing lift, and delaying stall. In order to improve the mission adaptability of small-scale UAVs, this dissertation focuses on the design and evaluation of the covert-inspired flow control device consisting of single and multiple rows of flaps for lift enhancement and stall mitigation. The effects of a single static covert-inspired flap during post-stall were first characterized. Specifically, the single flap effects on the aerodynamic forces for various single flap designs, i.e., flap locations, flap deflection angles, and flap mobility, were identified, and the flap effects on various baseline airfoils were explored at Reynolds numbers in the order of 10^5,where small scale UAVs operate. Detailed experiments and simulations were used to investigate how the covert-inspired flap affects lift and drag on an airfoil that exhibits sharp or sudden stall (i.e., the NACA 2414 airfoil) and one that exhibits soft or gradual stall (i.e., an E387(A) airfoil). The effects of the flap chord-wise locations and deflection angles on lift and drag were investigated through wind tunnel experiments for two types of flaps, namely a freely-moving flap and a static flap. Results show that the single static covert-inspired flap can delay stall by up to 5 deg and improve post-stall lift by up to 23%. However, the post-stall lift improvement characteristics and sensitivities are highly affected by the airfoil choice. For the soft stall airfoil (i.e., E387(A)), the stall onset delay is insensitive to changing the flap deflection angle, and the flap becomes ineffective when the flap location is changed. In contrast, for the sharp stall airfoil (i.e., NACA 2414), the post-stall lift improvements can be tuned using the flap deflection angle, and the flap remains effective over a wide range of chord-wise locations. Numerical studies reveal that the lift improvements are attributed to a step in the pressure distribution over the airfoil, which allows for lower pressures on the suction side upstream of the flap. These lower pressure regions on the airfoil with the flap are a result of the leading-edge vortices being closer to the airfoil surface when compared to the baseline, which increases the velocity over the suction side of the airfoil. The distinctions between the flap-induced lift enhancements on the soft and sharp stall airfoils suggest that the flap can be used as a tunable flow control device for the sharp stall airfoil, while for the soft stall airfoil, it can solely be used as a stall mitigation device that is either on or off.
In an effort to design a tunable flow control device, the remainder of the dissertation focused on the sharp stall airfoil. More specifically, the aerodynamic effects and dynamics of multiple freely-moving covert-inspired flaps were characterized on the NACA 2414 airfoil, which is the sharp stall airfoil. A simplified engineering analogy of the covert inspired flow control devices with multiple flaps was created, which helped to explore the feasibility of the multiple-flap designs and helped to formulate the first hypothesis about why birds deploy multiple rows of covert feathers during high-AoA flight. Detailed wind tunnel experiments were conducted, and the results showed that the multiple covert inspired flaps have additive, thus higher, post-stall lift improvement compared to the single flap systems. The additive effects of the flap-induced lift improvements on an engineering wing section were shown for the first time. Hot wire anemometry along with numerical simulations were carried out to study the effects of the multi-flap system on the flow field, including the turbulence wake profile, velocity fluctuation, pressure distribution, and flow visualization. Moreover, the flap structural dynamics were also studied for the first time. In the post-stall regime, a flap's deflection remains the same regardless of whether other flaps are deployed or not. Data-driven model of the multi-flap system was developed to create a framework that facilitates designing a system of covert-inspired flaps for various flight conditions. The detailed aerodynamic characterization and data-driven modeling efforts of this work enable the design of multiple spatially distributed flow control devices on engineering airfoils to achieve tunable post-stall lift improvement.Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2023-08-01The student, Chengfang Duan, accepted the attached license on 2021-07-14 at 17:55.The student, Chengfang Duan, submitted this Dissertation for approval on 2021-07-14 at 18:04.This Dissertation was approved for publication on 2021-07-15 at 17:58.DSpace SAF Submission Ingestion Package generated from Vireo submission #16945 on 2022-01-12 at 12:55:14Made available in DSpace on 2022-01-12T22:35:17Z (GMT). No. of bitstreams: 2
DUAN-DISSERTATION-2021.pdf: 22128087 bytes, checksum: 24f00761c4ef7d6143f5ac19838df306 (MD5)
LICENSE.txt: 4211 bytes, checksum: 6067f99d4e6d6bc116acd955a74b3669 (MD5)
Previous issue date: 2021-07-15Embargo set by: Seth Robbins for item 121127
Lift date: 2024-01-12T22:35:30Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemAuthor requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemU of I Onl
Local Public Goods Provision in the Post-Agricultural Tax Era in Rural China
No full textThis paper investigates regional differences in local public goods provision in rural area in the 2000s, using large village sample surveys (CHIP 2002 and 2007 surveys, a survey in Ningxia). Focuses are on changes in the coverage of public investment projects, regional differences in the determinants of public investment projects, and changes in the coverage of public services provided by village collectives. The main findings are as follows. First, we confirmed that coverage of public investment projects had increased in the 2000s. Second, in spite of concentration of fiscal administration into county level as one of the pillars of the reform of taxation and local fiscal system, administrative villages still played indispensable roles in local public goods provision. Third, we found that incentive of peasants, financial ability of villages, and incentive of local government affect location decision and budget structure of public investment projects and that direction and strength of such factors were different by regions.Local Public goods, Village, Local Government, Rural China
The Human Activity Radar Challenge: Benchmarking based on the ‘Radar signatures of human activities’ dataset from Glasgow University
No full textGreen Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Microwave Sensing, Signals & System
Integrated impact of urban mixed land use on TOD ridership: A multi-radius comparative analysis
No full textThe global trend toward urbanization has spurred the widespread adoption of transit-oriented development (TOD). While previous research has extensively explored the relationship between land use and TOD ridership, much of it has focused on linear associations at a singular scale. Leveraging recent advancements in nonlinear modeling and the accessibility of open-source data, this study employs a comprehensive two-step methodology. Firstly, K-means clustering algorithm categorizes TOD sites in Shenzhen into three distinct clusters, providing a site-based understanding of their characteristics. Subsequently, a Light Gradient Boosting Machine (LightGBM) classification model, complemented by SHapley Additive exPlanations (SHAP) values for interpretation, quantitatively evaluates the influence of mixed land use on TOD ridership across various catchment areas. As for the findings, we discover that land-use factors have different effects on TOD site patronage at different buffer radii and delve into the intricacies of these effects. Further results reveal non-linear relationships with varying degrees of positivity and negativity. For instance, residents and health sites positively impact patronage across all buffer radii, while certain commercial land uses exhibit a negative influence. The study demonstrates how the importance of different land-use structures varies across these clusters, shedding light on the nuanced impacts of land use on TOD catchment areas. Our research optimizes land-use mixes based on predominant cluster characteristics by offering actionable recommendations for urban managers
Crystal Ball: Prediction-based Real-time Evasion Against Deep Reinforcement Learning
No full textInternational audienc
Managing Large Multidimensional Array Hydrologic Datasets: A Case Study Comparing NetCDF and SciDB
No full textAbstractManagement of large hydrologic datasets including storage, structuring, indexing and query is one of the crucial challenges in the era of big data. This research originates from a specific data query problem: time series extraction at specific locations takes a long time when a large multidimensional dataset is stored in non-chunked NetCDF classic or 64-bit offset format. The essence of this issue lies in the contiguous storage structure adopted by NetCDF. In this research, NetCDF file based solutions and a multidimensional (MD) array database management system (DBMS) applying chunked storage structure are benchmarked to determine the best solution for storing and querying large hydrologic datasets. To achieve this, expert consultancy was conducted to establish benchmark sets. To guarantee a fair benchmark test environment, HydroNET-4 system was utilized and adapters for NetCDF files and SciDB were developed to manage and query data. In final benchmark tests, effect of data storage configurations such as chunk size and compression on query performance is also explored. Results indicate that SciDB arrays utilizing small chunk sizes show favorable performance. However with current implementation of SciDB, large numbers of small chunks cause huge overload of main memory which constraints SciDB's scalability. Compression of SciDB can either have negative or no effect on query performance, while it causes significant query degradation to NetCDF-4 solution. The research illustrates that for big hydrologic array data management, the properly chunked NetCDF-4 solution without compression is in general more efficient than the SciDB DBMS. So under current big data environment, traditionally adopted file-based hydroinformatic solutions can still be applicable after proper updating
- …
