Sponsor: Einride Autonomous Technologies US, Inc.
PI: Prof. Junfeng Zhao
The Phoenix metropolitan area has emerged as one of the nation’s most active ecosystems for autonomous vehicle testing and commercial deployment, particularly for robotaxi operations. Autonomous freight deployment along Arizona’s rural interstate corridors presents the State another opportunity and a fundamentally different autonomy challenge. Unlike urban robotaxis operating in dense environments with lower speeds and good road infrastructure, autonomous freight trucks must sustain long-duration, high-speed operation across sparsely instrumented desert corridors under extreme environmental exposure. Freight platforms introduce distinct safety and operational considerations such as higher vehicle mass, longer stopping distances, trailer dynamics, and logistics-driven uptime requirements. Einride is a global leader in autonomous trucking and plans to deploy autonomous electric freight operations along Arizona’s rural freight corridors. These corridors expose autonomous systems to sustained desert stressors that differ fundamentally from the rain, fog, and snow scenarios emphasized in most AV robustness studies. Extreme heat produces thermal shimmer and road mirage effects that distort lane geometry and generate false distant targets in camera systems. Elevated ambient infrared radiation increases the LiDAR receiver noise floor, reducing signal-to-noise ratio (SNR) and degrading effective detection range. Airborne dust can produce phantom LiDAR returns, false obstacle detections, and intermittent visibility degradation. Solar glare at low sun angles saturates forward-facing cameras. Critically, these stressors frequently co-occur in Arizona desert environments, producing compound and potentially non-linear degradation effects that are not represented in existing AV validation frameworks.
Sponsor: Arizona Department of Transportation (ADOT)/AZTI with funding from the Federal Highway Administration (FHWA) State Planning and Research (SPR) Program
PI: Dr. Junfeng Zhao; co-PI: Dr. Jeffrey Wishart, Dr. Brendan Russo, and Dr. Qiaoning Carol Zhang
Distracted driving remains a persistent roadway safety challenge in Arizona. ADOT reported
distracted driving being a factor in 3.5% (8,289 crashes) of the total number of crashes (236,193) in 20241; while the distraction factor was unknown in an additional 19.1% (44,990 crashes), suggesting that the true extent of distracted driving is likely higher than reported. As vehicle technology evolves, distraction is no longer limited to handheld phone use. Drivers also interact with in-vehicle infotainment systems, active safety, driving support-manual driver distraction guidelines emphasize that secondary tasks requiring drivers to look away from the roadway, manipulate controls, or monitor visual feedback can interfere with safe driving. The project will provide ADOT with evidence-based guidance for selecting distracted-driving technologies that balance safety, privacy, public acceptance, and implementation feasibility.
Sponsor: Arizona Department of Transportation (ADOT)/AZTI with funding from the Federal Highway Administration (FHWA) State Planning and Research (SPR) Program
PI: Dr. Junfeng Zhao; co-PI: Dr. Jeffrey Wishart, Dr. Brendan Russo, and Dr. Jun Zhao
The objective of the Zero-Emission Vehicle Crash Management System (ZEV-CMS) Mission is the development and of a system that will provide training, guidance, and tools for first and second responders to use when responding to a crash involving a ZEV (electric vehicle (EV) or fuel cell vehicle (FCV)). As ZEVs become more common on public roads, it is imperative that these responders understand the unique risks and challenges posed by these vehicles.
Sponsor: Arizona Department of Transportation (ADOT)/AZTI
with funding from the Federal Highway Administration (FHWA) State Planning and Research (SPR) Program
PI: Dr. Minfeng Shang and Dr. Jun Zhao; co-PI: Prof. Junfeng Zhao, Prof. Yao-Jan Wu
The work plan for this project is designed to support the Arizona Department of Transportation (ADOT) in evaluating and modernizing pavement marking practices to meet the evolving needs of both human drivers and advanced driver-assist vehicle technologies. As reliance on systems such as Lane Departure Warning (LDW), Lane Keeping Assistance (LKA), and Lane Centering Assistance (LCA) continues to grow, the visibility and consistency of pavement markings—particularly under challenging conditions such as nighttime, bad weather, and surface wear—has become increasingly important. This research aims to assess the adequacy of current ADOT pavement marking standards, compare them with best practices from other states, and develop recommendations to ensure compatibility with both legacy road users
and emerging vehicle technologies. The findings will guide the development of a statewide assessment program and a long-term maintenance strategy that promotes safer and more reliable roadway operations.
PI: Prof. Junfeng Zhao
We will develop a unified Transformer-based vehicle-to-infrastructure cooperative perception framework that enhances environmental understanding for connected and automated vehicles under occlusion, range limitation, and asynchronous sensing. VI-BEV introduces an intermediate fusion architecture in the Bird’s Eye View domain, leveraging cross-attention Transformers to integrate vehicle and infrastructure camera and LiDAR features for robust 3D object detection with extended perception range. Building on this representation, VI-Track extends cooperative perception to spatiotemporal reasoning by incorporating infrastructure assisted multi-object tracking and dynamics-aware fusion, enabling consistent object state estimation despite sensing and communication delays. Complementary to detection and tracking, VI-BEVSEG generalizes the same BEV-centric, attention-based fusion paradigm to cooperative semantic segmentation, supporting scene understanding tasks such as road layout and object occupancy mapping. Collectively, these methods establish a scalable, task-agnostic vehicle infrastructure perception framework that unifies detection, tracking, and segmentation through shared BEV representations and Transformer-based fusion, demonstrating improved robustness, coverage, and temporal consistency for safety-critical CAV applications.
Sponsor: BlackTeal Energy, Greater Phoenix Economic Council
PI: Prof. Junfeng Zhao
The integration of AI and Digital Twin technologies into Battery Management Systems (BMS) represents a transformative shift in energy storage management. These cutting-edge solutions address key challenges faced by traditional BMS, such as inefficiencies in state estimation, data fragmentation, and thermal management issues. By incorporating AI and Digital Twins, the proposed Unified Battery Management System (U-BMS) framework significantly enhances system performance, extends battery life, and improves compatibility and safety.
Sponsor: Arizona Commerce Authority, Science Foundation of Arizona
PI: Prof. Junfeng Zhao
The industry has progressed to a point where automated vehicles (AVs) are being deployed in limited commercial applications in select locations, including Arizona. However, whether the AV is commercially deployed or is a prototype being tested on public roads, the operational safety is unknown. Arizona State University, in collaboration with Science Foundation Arizona (SFAz), is developing an operational safety assessment (OSA) method to address this critical gap in public safety related to CAVs. The OSA method includes a mix of simulation, closed course, and public road scenario-based testing in order to assess the operational safety of the vehicle in thousands of scenarios, including difficult edge cases. The OSA method includes a mix of simulation, closed course, and public road scenario-based testing in order to assess the operational safety of the vehicle in thousands of scenarios, including difficult edge cases. The objective of the proposed project is to develop a platform that (1) enables sufficiently accurate simulation results through the creation of a high-fidelity model, known as a Digital Twin (DT), of a research CAV purchased by the ASU PI and (2) ensures more edge cases are encountered in closed course and public road testing by adding Augmented Reality (AR) capability to the research CAV. The proposed platform will allow for thousands of scenarios to be tested in simulation much faster and more cheaply than with the research CAV, which allows the latter to test a subset of scenarios for validation of the simulation results purposes, some of which will have AR-injected objects added to the research vehicle’s environment.
Sponsor: Cox Communications, Inc., ASU-Cox Collaboratory
PI: Prof. Junfeng Zhao
Finding an open parking space at congested parking lots in urban centers, commercial areas, and public spaces is often time-consuming and frustrating, leading to wasted fuel, and increased greenhouse gas emissions. Another concern in parking lots is the occurrence of traffic accidents, often caused by distracted drivers searching for parking spaces or navigating through congested lots. It is challenging not just for human drivers but also for Robotaxi or vehicles that can be summoned to parking lots. We need an innovative navigation solution to ease the parking process for human drivers and autonomous vehicles (AVs), and make it safer, more convenient, and environmentally friendly.
