conference paper

EcoFusion: energy-aware adaptive sensor fusion for efficient autonomous vehicle perception

Proceedings of the 59th ACM/IEEE Design Automation Conference

Publication Date

August 23, 2022

Author(s)

Arnav Vaibhav Malawade, Trier Mortlock, Mohammad Al Faruque

View conference paper

Abstract

Autonomous vehicles use multiple sensors, large deep-learning models, and powerful hardware platforms to perceive the environment and navigate safely. In many contexts, some sensing modalities negatively impact perception while increasing energy consumption. We propose EcoFusion: an energy-aware sensor fusion approach that uses context to adapt the fusion method and reduce energy consumption without affecting perception performance. EcoFusion performs up to 9.5% better at object detection than existing fusion methods with approximately 60% less energy and 58% lower latency on the industry-standard Nvidia Drive PX2 hardware platform. We also propose several context-identification strategies, implement a joint optimization between energy and performance, and present scenario-specific results.

Suggested Citation

Arnav Vaibhav Malawade, Trier Mortlock and Mohammad Abdullah Al Faruque (2022) “EcoFusion: energy-aware adaptive sensor fusion for efficient autonomous vehicle perception”, in Proceedings of the 59th ACM/IEEE Design Automation Conference. New York, NY, USA: Association for Computing Machinery (DAC '22), pp. 481–486. Available at: 10.1145/3489517.3530489.

conference paper

Can We Trust Embodied Agents? Exploring Backdoor Attacks against Embodied LLM-based Decision-Making Systems

The Thirteenth International Conference on Learning Representations (ICLR) 2025

Publication Date

April 30, 2025

Author(s)

Ruochen Jiao, Shaoyuan Xie, Justin Yue, Takami Sato, Lei Wang, Yu-Han (Doris) Wang, Qi Alfred Chen, Qi Zhu

View conference paper

Abstract

Large Language Models (LLMs) have shown significant promise in real-world decision-making tasks for embodied artificial intelligence, especially when fine-tuned to leverage their inherent common sense and reasoning abilities while being tailored to specific applications. However, this fine-tuning process introduces considerable safety and security vulnerabilities, especially in safety-critical cyber-physical systems. In this work, we propose the first comprehensive framework for Backdoor Attacks against LLM-based Decision-making systems (BALD) in embodied AI, systematically exploring the attack surfaces and trigger mechanisms. Specifically, we propose three distinct attack mechanisms: word injection, scenario manipulation, and knowledge injection, targeting various components in the LLM-based decision-making pipeline. We perform extensive experiments on representative LLMs (GPT-3.5, LLaMA2, PaLM2) in autonomous driving and home robot tasks, demonstrating the effectiveness and stealthiness of our backdoor triggers across various attack channels, with cases like vehicles accelerating toward obstacles and robots placing knives on beds. Our word and knowledge injection attacks achieve nearly 100% success rate across multiple models and datasets while requiring only limited access to the system. Our scenario manipulation attack yields success rates exceeding 65%, reaching up to 90%, and does not require any runtime system intrusion. We also assess the robustness of these attacks against defenses, revealing their resilience. Our findings highlight critical security vulnerabilities in embodied LLM systems and emphasize the urgent need for safeguarding these systems to mitigate potential risks.

Suggested Citation

Ruochen Jiao, Shaoyuan Xie, Justin Yue, Takami Sato, Lixu Wang, Yixuan Wang, Qi Alfred Chen and Qi Zhu (2025) “Can We Trust Embodied Agents? Exploring Backdoor Attacks against Embodied LLM-based Decision-Making Systems”, in The Thirteenth International Conference on Learning Representations (ICLR) 2025. Available at: https://ics.uci.edu/~alfchen/pubs/shaoyuan_iclr25.pdf (Accessed: August 21, 2025).

book/book chapter

Scene-Graph Embedding for Robust Autonomous Vehicle Perception

Publication Date

January 1, 2023

Author(s)

Shih-Yuan Yu, Arnav Vaibhav Malawade, Mohammad Al Faruque

View book/book chapter

Abstract

Robust Perception is vital in automotive Cyber-Physical Systems (CPS). Although the supporting technologies have advanced recently, enabling robust perception remains challenging for researchers and industry alike. The highly variable scenarios in complex urban environments can lead to erroneous perceptions, which are factors in most driver-related crashes. In this chapter, we present our experience developing AV perception models capable of better understanding driving scenes, thus improving their robustness. Specifically, we propose using scene-graphs as a better Intermediate Representation (IR) for road scenes. Besides, we develop a novel spatio-temporal graph learning approach based on scene-graph representations for modeling the risk of driving maneuvers. Our approach better understands driving scenes and converts them into an estimated risk level by leveraging a network architecture consisting of a Multi-Relation Graph Convolution Network (MR-GCN), a Long-Short Term Memory Network (LSTM), and self-attention layers. We demonstrate how a scene-graph approach for AV perception enables the AV to better assess risk across various driving maneuvers than state of the art, thus being more robust. Moreover, our approach can more effectively transfer knowledge learned from simulated data to real-world driving scenarios. Lastly, we show how adding spatial and temporal attention layers to our approach improves its explainability.

Suggested Citation

Shih-Yuan Yu, Arnav Vaibhav Malawade and Mohammad Abdullah Al Faruque (2023) “Scene-Graph Embedding for Robust Autonomous Vehicle Perception”, in V.K. Kukkala and S. Pasricha (eds.) Machine Learning and Optimization Techniques for Automotive Cyber-Physical Systems. Cham: Springer International Publishing, pp. 525–544. Available at: https://doi.org/10.1007/978-3-031-28016-0_18 (Accessed: October 23, 2024).

conference paper

Eve, You Shall Not Get Access! A Cyber-Physical Blockchain Architecture for Electronic Toll Collection Security

2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC)

Publication Date

September 1, 2020

Author(s)

Ahmed Didouh, Anthony Lopez, Yassin El Hillali, Atika Rivenq, Mohammad Al Faruque

View conference paper

Abstract

Cooperative intelligent transportation system (C-ITS) applications are generally susceptible to position spoofing-dependent attacks such as Sybil and DDoS attacks due to a lack of established solutions. This paper presents a novel cyber-physical blockchain cryptographic architecture to help prevent position spoofing attackers from becoming validated nodes in C-ITS applications. The solution also guarantees security requirements including the non-trivial non-repudiation in light of these and other attacks. With a use case of electronic toll collection (ETC), our architecture implements techniques based on Received Signal Strength Indication (RSSI) measurements in conjunction with blockchain authentication methods such as Proof-of-Location and smart contracts to determine the legitimacy of a node. We demonstrate our solution in experiments using ITS-G5 Cohda Wireless technology (a Road Side Unit and two On-Board Units programmed with the ITS Vanetza stack) with functionalities specified by the European Telecommunications Standardization Institute (ETSI). From our experimental results from several driving-based data gathering tests, we discovered that our solution is able to cope with noise and relative velocity challenges because it incorporates both OBUs and RSUs in the Proof of Location computation steps. In light of this, the proposed architecture may also be applicable to govern V2X in general.

Suggested Citation

Ahmed Didouh, Anthony Bahadir Lopez, Yassin El Hillali, Atika Rivenq and Mohammad Abdullah Al Faruque (2020) “Eve, You Shall Not Get Access! A Cyber-Physical Blockchain Architecture for Electronic Toll Collection Security”, in 2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC). 2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC), pp. 1–7. Available at: 10.1109/ITSC45102.2020.9294334.

published journal article

Evaluation and modification of constant volume sampler based procedure for plug-in hybrid electric vehicle testing

SAE Int. J. Alt. Power.

Publication Date

August 1, 2011

Author(s)

Li Zhang, Tim Brown, Scott Samuelsen

View published journal article

Suggested Citation

Li Zhang, Tim Brown and G. Scott Samuelsen (2011) “Evaluation and modification of constant volume sampler based procedure for plug-in hybrid electric vehicle testing”, SAE Int. J. Alt. Power., 1(2), pp. 542–559. Available at: 10.4271/2011-01-1750.

conference paper

Behavioral model for Ingress/Egress decision on buffer-separated HOV facilities for microsimulation model

Proceedings of the 89th annual meeting of the transportation research board

Publication Date

January 1, 2010

Author(s)

Shin-Ting (Cindy) Jeng, Will Recker, Lianyu Chu

Abstract

High-Occupancy Vehicle (HOV) lanes, commonly known as carpool lanes, have been accepted as a cost-effective and environmental-friendly alternative in many metropolitan areas. It is thus very important to find ways to evaluate HOV strategies during the decision making process, and prior to any implementation. An alternative approach is microscopic traffic simulation. However, current micro-simulation models have limited capabilities to model HOV driver behaviors, particularly with respect to buffer-separated facilities. This study proposed a probabilistic HOV driver behavioral model based on discrete choice modeling derived from random utility principles, which includes a preferred access choice model for examining travel time savings and a traffic model for calculating the acceptable gap to get in/out the HOV lane. A HOV access plugin was also developed using a micro-simulation tool, Paramics, based on the proposed model. Its theoretical capability is extended to real-world application via providing additional â??implementationâ?? parameters, such as ingress and egress points selection control, and update frequency of traffic information. A real-world freeway network, SR-57 in Orange County, California, was selected to analyze the reasonableness of the model through sensitivity analysis, and was further investigated for model validation purposes. The results have shown the reasonableness of the proposed model under various traffic conditions. The proposed model also demonstrated its feasibility and applicability via setting various calibration parameters and control parameters.

Suggested Citation

Shin-Ting (Cindy) Jeng, Will Recker and Lianyu Chu (2010) “Behavioral model for Ingress/Egress decision on buffer-separated HOV facilities for microsimulation model”, in Proceedings of the 89th annual meeting of the transportation research board, p. 32p.

Phd Dissertation

Exploring Delivery Service Substitution of Travel: Optimized Fleet Systems and Household Activity Patterns

Publication Date

January 1, 2024

Author(s)

Marjan Mosslemi

Areas of Expertise

Travel Behavior, Land Use, & the Built Environment

View Phd Dissertation

Abstract

This dissertation delves into the intersection of two critical elements shaping the future of transportation: opportunities and the challenges presented by shopping delivery services, particularly same-day delivery (SDD), and the necessity to anticipate and explore the forthcoming transportation paradigm with the new possibilities offered by Autonomous Vehicles (AVs). This study investigates the transformative potential of SDD services facilitated by a fleet of shared autonomous vehicles (SAVs) to reshape daily shopping trips and activities.With a dual focus on both the network and household layers, the dissertation addresses the viability of SDD services, considering vehicle miles traveled (VMT) savings and operational strategies for efficient fleet management on one side, and the impacts on travel patterns on the other. Leveraging real-world data for the network of Irvine, CA, and employing optimization methodologies, this dissertation (i) investigates the potential VMT savings from SDD compared to the base scenario where households conduct their own shopping activities, (ii) analyzes the optimal fleet size needed to achieve significant VMT reductions, and (iii) evaluates operational strategies for cost-effective and efficient service delivery. In this dissertation, I analyze the optimal fleet size and system design settings needed to achieve significant VMT reductions without losing profitability and I evaluate operational strategies for cost-effective and time-sensitive service delivery. At the network layer, the system is modeled as a multi-Vehicle and Multi-Depot Pickup and Delivery Problem with Time Windows (m-MDPDPTW), which was implemented in Google OR-Tools. The depots are assumed to be at the warehouse locations from where shopping goods deliveries are made. An analysis is presented for a delivery service comprising an AV fleet serving households on their daily shopping trips for the case study of the City of Irvine, CA. The results indicate these services can significantly decrease the distance traveled and the time spent for shopping trips. The dissertation tests several scenarios to determine how varying possible service operation parameters as well as demand characteristics would affect the results. Scenarios involving varying percentage of the service demand, time window for deliveries, loading/unloading time, and warehouse distribution are considered. At the household layer, the dissertation examines how the SDD service influences household travel patterns and savings, using output from the California Statewide Travel Demand Model (CSTDM) for the City of Irvine. The time saved is used as an accessibility measure. Using the Household Activity Travel Pattern Problem (HAPP), formulated as a pickup and delivery problem with time windows for household daily activities, time saved is compared over four distinct scenarios: a base (existing) case with CSTDM patterns, the HAPP-optimized version of the base case, the base case excluding shopping trips, and its HAPP-optimized version. HAPP-based analysis sheds light on new opportunities in travel and activity planning enabled by AVs as well as insights into future activity patterns shaped by subscription services that may lead to more optimized travel patterns. High Performance Computing is used to tackle the NP-Hard computational problem involved in HAPP in the real world case study with a large set of households. This research is also intended to establish the viability of operationalizing a HAPP-methodology for analyzing realistic travel network contexts, for transportation policies that involve innovative vehicle usage and routing patterns. A HAPP solution is not a model for the actual household-level travel behavior, but rather a constraint-driven optimal version of it. Nonetheless, with the availability of rich individual level activity data now and in the future, HAPP can indeed become an optimizer for households, if computational problems can be surmounted. This dissertation establishes that computational problems are not insurmountable with current cloud and advanced computing options, even for 4-member households with activities substitutable across individuals, which past research had generally avoided. The research illustrated that, for a real-world network that has an individual and household-level activity-based planning model, or at least a synthesized model of that kind, policy analysis for future transportation options can be done using HAPP to find an optimized implementation of the policy when the behavioral response to such policy is not available in the existing activity models or data. The dissertation also points to future research possibilities involving faster optimizations that can be achieved if HAPP can be implemented with starting feasible solutions that may be developed from existing networks.

Suggested Citation

Marjan Mosslemi (2024) Exploring Delivery Service Substitution of Travel: Optimized Fleet Systems and Household Activity Patterns. Ph.D.. UC Irvine. Available at: https://uci.primo.exlibrisgroup.com/permalink/01CDL_IRV_INST/u4evf/cdi_proquest_journals_3087787095 (Accessed: October 23, 2024).

conference paper

Revisiting Physical-World Adversarial Attack on Traffic Sign Recognition: A Commercial Systems Perspective

ISOC Network and Distributed System Security Symposium (NDSS) 2025

Publication Date

February 1, 2025

Author(s)

Ningfei Wang, Shaoyuan Xie, Takami Sato, Yunpeng Luo, Kaidi Xu, Qi Alfred Chen

View conference paper

Abstract

Traffic Sign Recognition (TSR) is crucial for safe and correct driving automation. Recent works revealed a general vulnerability of TSR models to physical-world adversarial attacks, which can be low-cost, highly deployable, and capable of causing severe attack effects such as hiding a critical traffic sign or spoofing a fake one. However, so far existing works generally only considered evaluating the attack effects on academic TSR models, leaving the impacts of such attacks on real-world commercial TSR systems largely unclear. In this paper, we conduct the first large-scale measurement of physical-world adversarial attacks against commercial TSR systems. Our testing results reveal that it is possible for existing attack works from academia to have highly reliable (100%) attack success against certain commercial TSR system functionality, but such attack capabilities are not generalizable, leading to much lower-than-expected attack success rates overall. We find that one potential major factor is a spatial memorization design that commonly exists in today’s commercial TSR systems. We design new attack success metrics that can mathematically model the impacts of such design on the TSR system-level attack success, and use them to revisit existing attacks. Through these efforts, we uncover 7 novel observations, some of which directly challenge the observations or claims in prior works due to the introduction of the new metrics.

Suggested Citation

Ningfei Wang, Shaoyuan Xie, Takami Sato, Yunpeng Luo, Kaidi Xu and Qi Alfred Chen (2025) “Revisiting Physical-World Adversarial Attack on Traffic Sign Recognition: A Commercial Systems Perspective”, in ISOC Network and Distributed System Security Symposium (NDSS) 2025. Available at: https://ics.uci.edu/~alfchen/pubs/ningfei_ndss25.pdf.

policy brief

Free Transit Helps Students—But Car Access, Transit Quality, and Perceptions Still Shape Who Benefits

Publication Date

June 22, 2026

Author(s)

Farzana Khatun, Henry Bernal, David Brownstone

View policy brief

Abstract

Fare-free transit programs are increasingly used to increase ridership and reduce household transportation costs, especially for lower-income households. In California, where fewer than ten percent of school trips are made on school buses and many students travel long distances to reach school, transit costs and access can pose real barriers to education and after school opportunities. At the same time, fare-free programs require substantial and ongoing public investment, raising important questions about whether eliminating fares alone is enough to meaningfully change student travel behavior.

In October 2021, LA Metro—the largest transit agency in Los Angeles County—launched the GoPass pilot program in partnership with other transit operators and multiple school districts, providing free transit passes to K-14 students. To better understand how GoPass influences student travel, who participates, and why, the research team examined student travel behavior, perceptions, and program participation across two complementary studies. One focused on Los Angeles Community College District (LACCD), including a survey of 1,800 LACCD students, while the other focused on high school students in a predominantly low-income school district.

conference paper

Towards Driving-Oriented Metric for Lane Detection Models

Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

Publication Date

January 1, 2022

Author(s)

Takami Sato, Qi Alfred Chen

View conference paper

Suggested Citation

Takami Sato and Qi Alfred Chen (2022) “Towards Driving-Oriented Metric for Lane Detection Models”. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 17153–17162. Available at: https://openaccess.thecvf.com/content/CVPR2022/html/Sato_Towards_Driving-Oriented_Metric_for_Lane_Detection_Models_CVPR_2022_paper.html (Accessed: October 5, 2023).