<p dir="ltr">E-commerce, fueled by the digital revolution, has become a cornerstone of modern retail, driving demand for efficient last-mile logistics services. As online sales soar past $4 trillion, the need for streamlined, cost-effective delivery solutions is urgent, particularly in markets like China, where complex traffic conditions and high customer expectations complicate last-mile delivery. Autonomous driving technology offers a promising approach to meeting these challenges, enabling lower costs and improved delivery efficiency. However, cities such as Suzhou present unique obstacles for autonomous delivery vehicles (ADVs), with unpredictable traffic and diverse obstacles like pedestrians and bicycles. To tackle these issues, this research developed a high-capacity simulation platform capable of executing 300,000 scenarios weekly. It incorporates advanced routing algorithms, such as the Shortest Path Faster Algorithm (SPFA), and high-definition mapping (HDMap) for precise localization, supporting rigorous testing across varied urban logistics scenarios. The platform’s modular microservices architecture ensures scalability, enabling thorough validation of both software and hardware components in unmanned logistics vehicles. Findings demonstrate that the platform’s architecture, particularly its modular microservices and Protocol Buffers for data handling, optimizes the reliability and safety of autonomous systems in dense urban environments. Realistic scenario generation through SPFA routing and HDMap integration provides a robust environment for decision-making tests, contributing to enhanced operational stability and efficiency.</p><p dir="ltr">Practical Implications extend beyond autonomous driving, suggesting relevance to intelligent transportation systems, delivery drones, and smart cities. The platform’s high-throughput capacity underscores the importance of large-scale testing, enabling rapid development cycles with minimal dependence on real-world testing. This research provides a foundation for future improvements in simulation efficiency, scenario diversity, and applications across various sectors, paving the way for further advancements in autonomous technology.</p>

<p dir="ltr">Introduces a new framework and state-of-the-art algorithm in closed-loop prediction for motion planning under differential constraints. More specifically, this work introduces the idea of sampling on specific "sampling regions" rather than the entire workspace to speed-up the motion planning process by orders of magnitude.</p>

<p dir="ltr">As automated vehicle (AV) systems become increasingly more intelligent, understanding the complex interplay between drivers' trust in these systems and their resulting behavior is paramount for the successful integration of autonomous technologies into the transportation landscape. Currently, the effects of displaying AV system-certainty information, concerning its navigability around obstacles, on drivers' trust, decision-making, and behavior is underexplored. This thesis seeks to address this research gap and evaluate a set of dynamic and continuous human-machine interfaces (HMIs) that present self-assessed system-certainty information to drivers of AVs. A simulated driving study was conducted wherein participants were exposed to four different linear and curvilinear AV system-certainty patterns when their AV approached a construction zone. The certainty patterns represented the vehicle’s confidence in safely avoiding the construction. Using this information, drivers needed to decide whether or not to take over from the vehicle. The AV’s reliability and system-certainty were not directly proportional to one another. During the study, drivers' trust, workload, takeover decisions and performance, eye movement behavior, and heart?rate measures were captured to comprehensively understand of the factors influencing drivers' interactions with automated vehicles. Overall, participants took over in 41.3% of the drives. Results suggest that the communication of different system-certainty trends had a significant effect on drivers’ takeover response times and gaze behavior, but did not affect their trust in the system nor their workload. Ultimately, the results of this work can be used to inform the design of in vehicle interfaces in future autonomous vehicles, aiming to enhance safety and driver acceptance. By elucidating the intricate relationship between drivers' trust and behavior, this study provides valuable insights for both researchers and developers, contributing to the ongoing discourse on the human factors associated with the integration of autonomous technologies into the transportation ecosystem.</p>

<p dir="ltr">High-speed autonomous racing is a unique and challenging research field that requires sophisticated simulation tools for algorithm development and validation. The difficulty in simulation development lies in bridging the sim-to-real gap by accurately modeling the dynamics of a real-world vehicle, external forces, and simulating realistic sensor data. As a competitor in the Indy Autonomous Challenge (IAC), the Purdue AI Racing (PAIR) team has had to build and iteratively improve upon their simulation tools while meticulously validating its performance with real-world data.</p><p dir="ltr">This thesis provides an in-depth overview of existing autonomous vehicle simulators and details the design of the Purdue AI Racing Simulator (PAIRSim), a novel simulation platform for bridging software-in-the-loop (SIL) and hardware-in-the-loop (HIL) testing. PAIRSim has played a pivotal role in the Purdue AI Racing (PAIR) team's development pipeline and has contributed to the team's success in past IAC race seasons, including a podium finish in 2025. PAIRSim offers a range of sensor capabilities including navigational GNSS/IMU sensors, as well as perception sensors like camera and LiDAR. Furthermore, PAIRSim is a highly modular and user-friendly simulator that improves upon existing simulators in the field in both form and function. By using real-world data gathered during an IAC 2025 competition run as a means of simulator validation, it is clear that there is immense overlap between PAIRSim and reality. Data gathered from a simulation of the same competition run demonstrates that PAIRSim accurately replicates the dynamic behavior of a real-world racecar.</p><p dir="ltr">Therefore, the insights provided in this thesis will advance the field of high-speed autonomy by providing an entry-point into simulation for future teams participating in the IAC and for general users who want to test autonomous racing algorithms. </p>

<p dir="ltr">The growth of electric vehicles (EVs) offers several benefits for air quality improvement and emissions reduction. Nonetheless, EVs also pose several challenges in the area of highway transportation. These barriers are related to the limitations of EV technology, particularly the charge duration and speed of battery recharging, which translate to vehicle range anxiety for EV users. A promising solution to these concerns is V2V DWC technology (Vehicle to Vehicle Dynamic Wireless Charging), particularly mobile energy disseminators (MEDs). The MED is mounted on a large vehicle or truck that charges all participating EVs within a specified locus from the MED. However, current research on MEDs offers solutions that are widely considered impractical for deployment, particularly in urban environments where range anxiety is common. Acknowledging such gap in the literature, this thesis proposes a comprehensive methodological framework for optimal MED deployment decisions. In the first component of the framework, a practical system, termed “ChargingEnv” is developed using reinforcement learning (RL). ChargingEnv simulates the highway environment, which consists of streams of EVs and an MED. The simulation accounts for a possible misalignment of the charging panel and incorporates a realistic EV battery model. The second component of the framework uses multiple deep RL benchmark models that are trained in “ChargingEnv” to maximize EV service quality within limited charging resource constraints. In this study, numerical experiments were conducted to demonstrate the MED deployment decision framework’s efficacy. The findings indicate that the framework’s trained model can substantially improve EV travel range and alleviate battery depletion concerns. This could serve as a vital tool that allows public-sector road agencies or private-sector commercial entities to efficiently orchestrate MED deployments to maximize service cost-effectiveness.</p>

<p dir="ltr">State estimation is a type of an <i>inverse problem</i> in which some amount of observed data needs to be processed using computer algorithms (which are designed using analytical techniques) to infer or reconstruct the underlying model that produced the data. Due to the ubiquity of data and interconnected control systems in the present day, many engineering domains have become replete with inverse problems that can be formulated as state estimation problems. The interconnectedness of these control systems imparts the associated state estimation problems with distinctive structural properties that must be taken into consideration. For instance, the observed data could be high-dimensional and have a dependency structure that is best described by a graph. Furthermore, the control systems of today interface with each other and with the internet, bringing in new possibilities for large-scale collaborative sensor fusion, while also (potentially) introducing new sources of disturbances, faults, and cyberattacks. </p><p dir="ltr">The main thesis of this document is to investigate the unique challenges related to the issues of robustness, resilience (to faults and cyberattacks), and scalability of state estimation algorithms. These correspond to research questions such as, <i>"Does the state estimation algorithm retain its performance when the measurements are perturbed by unknown disturbances or adversarial inputs?"</i> and <i>"Does the algorithm have any bottlenecks that restrict the size/dimension of the problems that it could be applied to?".</i> Most of these research questions are motivated by a singular domain of application: autonomous navigation of unmanned aerial vehicles (UAVs). Nevertheless, the mathematical methods and research philosophy employed herein are quite general, making the results of this document applicable to a variety of engineering tasks, including anomaly detection in time-series data, autonomous remote sensing, traffic monitoring, coordinated motion of dynamical systems, and fault-diagnosis of wireless sensor networks (WSNs), among others.</p>

<p dir="ltr">Autonomous vehicles require motion forecasting of their surrounding multiagents (pedestrians</p><p dir="ltr">and vehicles) to make optimal decisions for navigation. The existing methods focus on</p><p dir="ltr">techniques to utilize the positions and velocities of these agents and fail to capture semantic</p><p dir="ltr">information from the scene. Moreover, to mitigate the increase in computational complexity</p><p dir="ltr">associated with the number of agents in the scene, some works leverage Euclidean distance to</p><p dir="ltr">prune far-away agents. However, distance-based metric alone is insufficient to select relevant</p><p dir="ltr">agents and accurately perform their predictions. To resolve these issues, we propose the</p><p dir="ltr">Semantics-aware Interactive Multiagent Motion Forecasting (SIMMF) method to capture</p><p dir="ltr">semantics along with spatial information and optimally select relevant agents for motion</p><p dir="ltr">prediction. Specifically, we achieve this by implementing a semantic-aware selection of relevant</p><p dir="ltr">agents from the scene and passing them through an attention mechanism to extract</p><p dir="ltr">global encodings. These encodings along with agents’ local information, are passed through</p><p dir="ltr">an encoder to obtain time-dependent latent variables for a motion policy predicting the future</p><p dir="ltr">trajectories. Our results show that the proposed approach outperforms state-of-the-art</p><p dir="ltr">baselines and provides more accurate and scene-consistent predictions. </p>

<p>Upon introducing semi- and fully automated vehicles on the road, drivers will be reluctant to focus on the traffic interaction and rely on the vehicles' decision-making. However, encountering pedestrians still poses a significant difficulty for modern automated driving technologies. Considering the high-level complexity in human behavior modeling to solve a real-world problem, deep-learning algorithms trained from naturalistic data have become promising solutions. Nevertheless, although developing such algorithms is achievable based on scene data collection and driver knowledge extraction, evaluation remains challenging due to the potential crash risks and limitations in acquiring ground-truth intention changes. </p> <p><br></p> <p>This study proposes a VR-based testing bed to evaluate real-time pedestrian intention algorithms as VR simulators are recognized for their affordability and adaptability in producing a variety of traffic situations, and it is more reliable to conduct human-factor research in autonomous cars. The pedestrian wears the head-mounted headset or uses the keyboard input and makes decisions in accordance with the circumstances. The simulator has added a credible and robust experience, essential for exhibiting the real-time behavior of the pedestrian. While crossing the road, there exists uncertainty associated with pedestrian intention. Our simulator will anticipate the crossing intention with consideration of the ambiguity of the pedestrian behavior. The case study has been performed over multiple subjects in several crossing conditions based on day-to-day life activities. It can be inferred from the study outcomes that the pedestrian intention can be precisely inferred using this VR-based simulator. However, depending on the speed of the car and the distance between the vehicle and the pedestrian, the accuracy of the prediction can differ considerably in some cases.</p>

<p dir="ltr">The future of the global economy lies in space. As the economic and scientific benefits from space become more accessible and apparent to the public, the demand for more spacecrafts will only increase. However, simply using the current space architecture to sustain any major activities past low Earth orbit is infeasible. The limiting factor of relying on ground operators via the Deep Space Network will blunt future growth in cislunar space traffic as the bandwidth is insufficient to satisfy the needs of every spacecraft in this domain. For this reason, spacecrafts must begin to operate autonomously or semi-autonomously for operators to be able to manage more missions at a given time. This thesis focuses on the guidance and navigation policies that could help vehicles such as logistical or resupply spacecrafts perform their rendezvous autonomously. It is found that using GNSS signals and Moon-based optical navigation has the potential to help spacecrafts perform autonomous orbit determination in near-Moon trajectories. The estimations are high enough quality such that a stochastic controller can use this navigation solution to confidently guide the spacecraft to a target within a tolerance before proximity operations commence. As the reliance on the ground is shifted away, spacecrafts would be able to operate in greater numbers outside of Earth's lower orbits, greatly assisting humanity's presence in space. </p>

<p dir="ltr">Traffic safety and congestion are global concerns. Autonomous vehicles (AVs) are expected to enhance transportation safety and reduce congestion. However, achieving their full potential requires 100% market penetration, a challenging task. This study addresses key issues in mixed traffic environments, where human-driven vehicles (HDVs) and connected autonomous vehicles (CAVs) coexist. A number of critical questions persist: 1) inadequate exploration of human errors (errors originating from non-CAV sources) in mixed traffic; 2): limited focus on information selection and learning efficiency in network-level rerouting, particularly in highly dynamic environments; 3) inadequacy of personalized element driver inputs in motion-planning frameworks; 4) lack of consideration of user privacy concerns.</p><p dir="ltr">With the goal of advancing the existing knowledge in this field and shedding light on these matters, this dissertation introduces multiple frameworks. These frameworks leverage connectivity and automation to improve safety and mobility in mixed traffic, addressing various research levels, including local-level and network-level safety enhancement, as well as network-level and global-level mobility enhancement. With optimization- and learning-based methods implemented (Model Predictive Control, Deep Neural Network, Deep Reinforcement Learning, Transformer model and Federated Learning), frameworks introduced in this dissertation are expected to help highway agencies and vehicle manufacturers improve the safety and efficiency of traffic flow in the mixed-traffic era. Our research findings revealed increased crash-avoidance rates in critical situations, enhanced accuracy in predicting lane changes, improved dynamic rerouting within urban areas, and the implementation of effective data-sharing mechanisms with a focus on user privacy. This research underscores the potential of connectivity and automation to significantly enhance mixed-traffic safety and mobility.</p>