Indiana University-Purdue University Indianapolis (IUPUI) / Autonomous vehicles currently dominate the automobile field for their impact on humanity and society. Connected and Automated Vehicles (CAV’s) are vehicles that use different communication technologies to communicate with other vehicles, infrastructure, the cloud, etc. With the information received from the sensors present, the vehicles analyze and take necessary steps for smooth, collision-free driving. This the sis talks about the cruise control system along with the intersection collision avoidance system based on Petri net models. It consists of two internal controllers for velocity and distance control, respectively, and three external ones for collision avoidance. Fault-tolerant redundant controllers are designed to keep these three controllers in check. The model is built using a PN toolbox and tested for various scenarios. The model is also validated, and its distinct properties are analyzed.

Indiana University-Purdue University Indianapolis (IUPUI) / The current technologies employed for autonomous driving provide tremendous performance and results, but the technology itself is far from mature and relatively expensive. Some of the most commonly used components for autonomous driving include LiDAR, cameras, radar, and ultrasonic sensors. Sensors like such are usually high-priced and often require a tremendous amount of computational power in order to process the gathered data. Many car manufacturers consider cameras to be a low-cost alternative to some other costly sensors, but camera based sensors alone are prone to fatal perception errors. In many cases, adverse weather and night-time conditions hinder the performance of some vision based sensors. In order for a sensor to be a reliable source of data, the difference between actual data values and measured or perceived values should be as low as possible. Lowering the number of sensors used provides more economic freedom to invest in the reliability of the components used. This thesis provides an alternative approach to the current autonomous driving methodologies by utilizing acoustic signatures of moving objects. This approach makes use of a microphone array to collect and process acoustic signatures captured for simultaneous localization and mapping (SLAM). Rather than using numerous sensors to gather information about the surroundings that are beyond the reach of the user, this method investigates the benefits of considering the sound waves of different objects around the host vehicle for SLAM. The components used in this model are cost-efficient and generate data that is easy to process without requiring high processing power. The results prove that there are benefits in pursuing this approach in terms of cost efficiency and low computational power. The functionality of the model is demonstrated using MATLAB for data collection and testing.

Real-time path generation, including collision checks, is vital in critical driving scenarios such as navigating non-signalized intersections. These intersections lack organized traffic flow, which raises the risk of accidents. Rapidly Exploring Random Trees (RRT) is a widely adopted algorithm in robotics for motion planning due to its simplicity and probabilistic completeness. Over the years, researchers have made modifications to the basic RRT algorithm to improve its performance in dynamic environments, making it a favored planning algorithm for autonomous driving. Among these variants, probabilistic RRT (pRRT) demonstrates promising capabilities for efficient online replanning. The first part of the thesis thoroughly studies the pRRT algorithm and compares its performance to the standard RRT and RRT* algorithms through Python simulations. The pRRT algorithm outperformed the RRT and RRT* algorithms in terms of success rate and time to find a safe trajectory. The algorithm was implemented experimentally on scaled cars for the validation of its feasibility. The experimental results show good sim-to-real transfer for this algorithm. The second part of the thesis proposes a novel algorithm for path planning. The algorithm outperforms the standard RRT and pRRT techniques in terms of optimality and conformance to human instincts. The generated paths are much smoother and easier for the controller to track. The AV implementation combines the probabilistic RRT with the RRT-Connect algorithm to mitigate the problem of parameter tuning of the standard pRRT algorithm. The idea is to generate intermediate critical points around the obstacles to grow multiple trees between these points, which are then eventually connected if a safe trajectory is found. The algorithm was tested in simulation and showed comparatively better performance in handling obstacles. / Master of Science / Due to uncontrolled traffic flow, non-signalized intersections are critical for autonomous driving. Motion planning is responsible for the vehicle's decision-making and generating actions based on its surroundings. Rapidly Exploring Random Trees (RRT) is one of the most widely used algorithms for motion planning in robotics due to its simplicity and a guarantee of finding a collision-free path if it exists. Due to the randomness of the algorithm, the time to find a collision-free path increases rapidly as the surrounding environment complicates. In this thesis, we thoroughly study a modified version of RRT called the probabilistic RRT (pRRT) for motion planning of autonomous vehicles. The pRRT algorithm reduces the randomness of the standard RRT algorithm and takes into account the destination location and the positions of the obstacles to find a path around the obstacles and toward the destination point. The algorithm was experimentally validated and confirmed the simplistic transfer from simulations to reality. In the second part of the thesis, we propose a novel algorithm that combines the properties of pRRT and another well-known algorithm called RRT-Connect. This algorithm plans collision-free paths from the start, and the goal points towards free space around the obstacles simultaneously and then combines these fragmented paths. This reduces the overall planning time and was found to be better at providing smooth paths.

No description available.

No description available.

No description available.

The use of connected and autonomous ground and aerial vehicles is a promising solution to reduce accidents, improve the traffic efficiency, and provide various services ranging from delivery of goods to monitoring. Different from the current connected vehicles and autonomous vehicles, connected and autonomous vehicles (CAVs) combine autonomy and wireless connectivity and use both sensors and communication systems to increase their situational awareness and for their decision-making. However, in order to reap all the benefits of deploying CAVs, one must consider the interconnection between communication, control, and learning mechanisms for the CAV system design. The key goal of this dissertation is, thus, to develop foundational science that can be used for the design, analysis, and optimization of CAV systems while jointly taking into account the synergies among communication, control, and learning systems. First, a joint communication and control system design is developed for non-coordinated CAVs when performing autonomous path tracking. In particular, the maximum time delay requirements are derived to guarantee the stability of the controller when tracking two typical road scenarios (i.e., straight line and circular curve). Tools from optimization theory and risk theory are then used to jointly optimize the control system and power allocation for the communication network so as to maximize the number of vehicular links that meet the controller's delay requirements. Second, the joint control and communication design framework is extended to two coordinated CAVs applications, i.e., CAV platoons and unmanned aerial vehicle (UAV) swarms. Third, a distributed machine learning algorithm, i.e., federated learning (FL), is proposed for a swarm of connected and autonomous UAVs to execute tasks, such as coordinated trajectory planning and cooperative target recognition. In particular, a rigorous convergence analysis for FL is performed to show how wireless factors impact the FL convergence performance, and the design of UAV swarm networks is optimized to reduce the convergence time. Fourth, a new FL framework, called dynamic federated proximal (DFP) algorithm, is proposed for designing the autonomous controller of CAVs while considering the mobility of CAVs, the wireless fading channels, as well as the unbalanced and non independent and identically distributed data across CAVs. To improve the convergence of the proposed DFP algorithm, a contract-theoretic incentive mechanism is also proposed. Fifth, a wireless-enabled asynchronous federated learning (AFL) framework is proposed for urban air mobility (UAM) aircraft to collaboratively learn the turbulence prediction model. In particular, to characterize how UAM aircraft leverage wireless connectivity for AFL, a stochastic geometry based spatial model is developed and the wireless connectivity performance is analyzed. Then, a rigorous convergence analysis is performed for the proposed AFL framework to identify how fast the UAM aircraft converge to using the optimal turbulence prediction model. Sixth, based on the concordance order from stochastic ordering theory, a dependence control mechanism is proposed to improve the overall reliability of wireless networks for CAVs. Finally, to determine the optimal cache placement for CAVs, a novel spatio-temporal caching framework is proposed where the notion of graph motifs, i.e., the spatio-temporal communication patterns in wireless networks, is used. In conclusion, the frameworks presented in this dissertation will provide key fundamental guidelines to design, analyze, and optimize CAV systems. / Doctor of Philosophy / The evolution of transportation systems has always been the key to the progress of human societies. Recently, technology advances in sensing, autonomy, computing, and wireless connectivity ushered in the era of connected and autonomous vehicles (CAVs). In essence, CAVs rely on the data collected from sensors and wireless communication systems to automatically make the operation decision. If designed properly, the deployment of CAVs can improve the safety and the driving experience, increase the fuel efficiency and road capacity, as well as provide various services ranging from delivery of goods to monitoring. To reap all these benefits of deploying CAVs, one must address a number of technique challenges related to the wireless connectivity, autonomy, and autonomous learning for CAV systems. In particular, for CAV connectivity, the challenges include building a low latency and highly reliable network, using proper models for mobile radio channels, and determining the effective content dissemination strategy. At the control level, key considerations include guaranteeing stability and robustness for the controller when faced with measurement errors and wireless imperfections and rapidly adapting the CAV to dynamic environments. Meanwhile, when CAVs use machine learning to complete their tasks (e.g., object detection and environment monitoring), insufficient training data, privacy concerns, communication overhead, and limited energy are among the main challenges. Therefore, this dissertation develops the foundational science needed to design, analyze, and optimize CAVs while jointly taking into account the challenges within the wireless network, controller, and leaning mechanism design. To this end, various frameworks for the joint communication, control, and learning design and wireless network optimizations are proposed for different CAV applications. The results show that, using the proposed frameworks, the performance of CAVs can be optimized with more reliable communication systems, more stable controller, and improved learning mechanism, enabling intelligent transportation systems for the future smart cities.

Tracking objects in the surrounding environment is a key component of safe navigation for autonomous vehicles. An accurate tracking algorithm is required following object identification and association. This thesis presents the design and implementation of an adaptive Kalman filter for tracking objects commonly observed by autonomous vehicles. The design results from an evaluation of motion models, noise assumptions, fast error convergence methods, and methods to adaptively compensate for unexpected object motion. Guidelines are provided on these topics. Evaluation is performed through Monte Carlo simulation and with real data from the KITTI autonomous vehicle benchmark. The adaptive Kalman filter designed is shown to be capable of accurately tracking both typical and harsh object motions. / Master of Science / Tracking surrounding objects is a key challenge for autonomous vehicles. After the type of object is identified, and it is associated as either a newly or previously observed object, it is useful to develop a mathematical model of where it may go next. The Kalman filter is an algorithm capable of being employed for this purpose. This thesis presents the design of a Kalman filter tuned for tracking objects commonly observed by autonomous vehicles and augmented to handle object motion exceeding its base design. The design results from an evaluation of relevant mathematical models of an object's motion, methods to quickly reduce the error of the filter's estimate, and methods to monitor the filter's performance to see if it is operating outside of normal bounds. Evaluation is performed through simulation and with real data from the KITTI autonomous vehicle benchmark. The adaptive Kalman filter designed is shown to be capable of accurately tracking both typical and harsh object motions.

The Autonomous Challenge component of the Intelligent Ground Vehicle Competition (IGVC) requires robots to autonomously navigate a complex obstacle course. The roadway-type course is bounded by solid and broken white and yellow lines. Along the course, the vehicle encounters obstacles, painted potholes, a ramp and a sand pit. The success of the robot is usually determined by the software controlling it. Johnny-5 was one of three vehicles entered in the 2004 competition by Virginia Tech. This paper presents the vision processing software created for Johnny-5. Using a single digital camera, the software must find the lines painted in the grass, and determine which direction the robot should move. The outdoor environment can make this task difficult, as the software must cope with changes in both lighting and grass appearance. The vision software on Johnny-5 starts by applying a brightest pixel threshold to reduce the image to points most likely to be part of a line. A Hough Transform is used to find the most dominant lines in the image and classify the orientation and quality of the lines. Once the lines have been extracted, the software applies a set of behavioral rules to the line information and passes a suggested heading to the obstacle avoidance software. The effectiveness of this behavior-based approach was demonstrated in many successful tests culminating with a first place finish in the Autonomous Challenge event and the $10,000 overall grand prize in the 2004 IGVC. / Master of Science

The Nemesis tracked vehicle platform is a differentially driven Humanitarian Demining tractor developed by Applied Research Associates, Inc. The vehicle is capable of teleoperational control and is outfitted with a sensor suite used for detecting and neutralizing landmines. Because the detection process requires the vehicle to travel at speeds less than 0.5 km/h, teleoperation is a tedious process. The added autonomous capabilities of waypoint navigation and obstacle avoidance could greatly reduce operator fatigue. ARA chose to leverage Virginia Tech's experience in developing an autonomous mobility capability for the Nemesis platform. The resulting algorithms utilize the waypoint navigation techniques of Virginia Tech's JAUS (Joint Architecture for Unmanned Systems) toolkit, and a modified version of the Dynamic Expanding Zones (DEZ) algorithm developed for the 2005 DARPA Grand Challenge. The modified approach discretizes the perception zones of the DEZ algorithm and provides the added capability of obstacle memory, resulting in the Discretized Dynamic Expanding Zones with Memory (DDEZm) algorithm. These additions are necessary for efficient autonomous control of the differentially driven Nemesis vehicle. The DDEZm algorithm was coded in LabVIEW and used to autonomously navigate the Nemesis vehicle through a waypoint course while avoiding obstacles. The Joint Architecture for Unmanned Systems (JAUS) was used as the communication standard to facilitate the interoperability between the software developed at Virginia Tech and the existing Nemesis software developed by ARA. In addition to development and deployment, the algorithm has been fully documented for embedded coding by a software engineer. With embedded implementation on the vehicle, this algorithm will help to increase the efficiency of the landmine detection process, ultimately saving lives. / Master of Science