Search for: All records

We analyse a suite of 29 high-resolution zoom-in cosmological hydrodynamic simulations of massive galaxies with stellar masses $M_{\rm star} \gt 10^{10.9} \, \mathrm{M}_\odot$, with the goal of better understanding merger activity among active galactic nuclei (AGN), AGN activity in merging systems, SMBH growth during mergers, and the role of gas content in triggering AGN. Using the radiative transfer code Powderday, we generate HST-WFC3 F160W mock observations of central galaxies at redshift 0.5 < z < 3; convolve each image with a CANDELS-like point spread function; stitch each image into a real CANDELS image; and identify mergers within the synthetic images using commonly adopted non-parametric statistics. We study the connection between mergers and AGN activity in both the simulations and synthetic images and find reasonable agreement with observations from CANDELS. We find that AGN activity is not primarily driven by major mergers (stellar mass ratio > 1:4) except in a select few cases of gas-rich mergers at low redshifts (0.5 < z < 0.9). We also find that major mergers do not significantly grow the central SMBHs, indicating major mergers do not sustain long-term accretion. Moreover, the most luminous AGN in our simulations (Lbol > 1045 erg s−1) are no more likely than inactive galaxies (Lbol < 1043 erg s−1) to be found in merging systems. We conclude that mergers are not the primary drivers of AGN activity in the simulated massive galaxies studied here.

 

Abstract—This Research Work In Progress Paper examines empirical evidence on the impacts of feedback from an intelligent tutoring software on sketching skill development. Sketching is a vital skill for engineering design, but sketching is only taught limitedly in engineering education. Teaching sketching usually involves one-on-one feedback which limits its application in large classrooms. To meet the demands of feedback for sketching instruction, SketchTivity was developed as an intelligent tutoring software. SketchTivity provides immediate personalized feedback on sketching freehand practice. The current study examines the effectiveness of the feedback of SketchTivity by comparing students practicing with the feedback and without. Students were evaluated on their motivation for practicing sketching, the development of their skills, and their perceptions of the software. This work in progress paper examines preliminary analysis in all three of these areas. 

Vehicles can utilize their sensors or receive messages from other vehicles to acquire information about the surrounding environments. However, the information may be inaccurate, faulty, or maliciously compromised due to sensor failures, communication faults, or security attacks. The goal of this work is to detect if a lane-changing decision and the sensed or received information are anomalous. We develop three anomaly detection approaches based on deep learning: a classifier approach, a predictor approach, and a hybrid approach combining the classifier and the predictor. All of them do not need anomalous data nor lateral features so that they can generally consider lane-changing decisions before the vehicles start moving along the lateral axis. They achieve at least 82% and up to 93% F1 scores against anomaly on data from Simulation of Urban MObility (SUMO) and HighD. We also examine system properties and verify that the detected anomaly includes more dangerous scenarios. 

In current practice, exploring the computation and software level of individual ECUs of an automotive system does not seem feasible enough for a system-level understanding of vehicular electronics. Exploring vehicular system-level use cases requires exercising the communication and coordination of the constituent ECUs. We are developing a prototype environment, VIVE, to enable early exploration of system-level coordination. VIVE enables extensible use case definition, as well as smooth and seamless addition of new, compute, sensor, or actuation functionality. This solution is flexible and configurable in such a way that enables the user to exercise inter-component and intersystem interactions. In this paper, we demonstrate the utility of such a prototyping environment in the exploration of a traction control use case. 

Freehand sketching equips engineers to represent ideas rapidly in the design process, but most engineering curriculums fall short of equipping students with adequate sketching skills. This paper is focused on methods to improve engineers’ sketching skill through type of instruction, length of instruction, and delivery of and feedback for assignments using Sketchtivity, an intelligent sketch-tutoring software. We answer several key questions for providing better sketching education for engineers. Does perspective training improve freehand drawing ability? Can an intelligent tutoring software improve education outcomes? And how much sketching instruction is necessary for engineers? Analyzing the changes in sketching skill from pre- to post-sketching instruction between different instruction types (n = 116), we found that perspective sketching instruction significantly improved freehand sketching ability compared to traditional engineering sketching methods. When comparing pre to post sketching skill of students using Sketchtivity (n = 135), there was no significant difference in improvement between students using the intelligent tutoring software and those that exclusively practiced on paper – both groups improved equally. However, completing sketching tasks on tablets did not hinder students’ skill development even when measured on paper. Future work will more directly explore the influence of Sketchtivity on sketching skill development. Additionally, we found that five weeks of sketching instruction greatly improves sketching skill compared to only three weeks of instruction (n = 108), but both approaches significantly improve sketching self-efficacy. These outcomes support more extensive sketching instruction in engineering classrooms, and changes in instruction type to promote more freehand sketching skills. 

Abstract—This Research Work In Progress Paper examines empirical evidence on the impacts of feedback from an intelligent tutoring software on sketching skill development. Sketching is a vital skill for engineering design, but sketching is only taught limitedly in engineering education. Teaching sketching usually involves one-on-one feedback which limits its application in large classrooms. To meet the demands of feedback for sketching instruction, SketchTivity was developed as an intelligent tutoring software. SketchTivity provides immediate personalized feedback on sketching freehand practice. The current study examines the effectiveness of the feedback of SketchTivity by comparing students practicing with the feedback and without. Students were evaluated on their motivation for practicing sketching, the development of their skills, and their perceptions of the software. This work in progress paper examines preliminary analysis in all three of these areas. 

We develop a virtual prototyping infrastructure for modeling and simulation of automotive systems. We focus on exercising and exploring use cases involving system-level coordination of vehicular electronics, sensors, and software. In current practice, such use cases can only be explored late in the design when all the relevant hardware components are available. Any design change, e.g., for optimization or security or even functional errors found during the exploration, incurs prohibitive cost at that stage. Our solution is a flexible, configurable prototyping platform that enables the user to seamlessly add new system-level use cases. Unlike other related prototyping environments, the focus of our platform is on communication and coordination among different components, not the computation of individual Electronic Control Units. We report on the use of the platform for implementing several realistic usage scenarios on automotive platforms and exploring the effects of their interaction. In particular, we show how to use the platform to develop real-time in-vehicle communication optimizers for different optimization targets. 

Connected Autonomous Vehicular (CAV) platoon refers to a group of vehicles that coordinate their movements and operate as a single unit. The vehicle at the head acts as the leader of the platoon and determines the course of the vehicles following it. The follower vehicles utilize Vehicle-to-Vehicle (V2V) communication and automated driving support systems to automatically maintain a small fixed distance between each other. Reliance on V2V communication exposes platoons to several possible malicious attacks which can compromise the safety, stability, and efficiency of the vehicles. We present a novel distributed resiliency architecture, RePLACe for CAV platoon vehicles to defend against adversaries corrupting V2V communication reporting preceding vehicle position. RePLACe is unique in that it can provide real-time defense against a spectrum of communication attacks. RePLACe provides systematic augmentation of a platoon controller architecture with real-time detection and mitigation functionality using machine learning. Unlike computationally intensive cryptographic solutions RePLACe accounts for the limited computation capabilities provided by automotive platforms as well as the real-time requirements of the application. Furthermore, unlike control-theoretic approaches, the same framework works against the broad spectrum of attacks. We also develop a systematic approach for evaluation of resiliency of CAV applications against V2V attacks. We perform extensive experimental evaluation to demonstrate the efficacy of RePLACe.