<div>The heavy duty transportation sector is projected to grow in the coming decades. Increasing the fuel economy of class 8 vehicles would simultaneously decrease CO2 emissions and decrease the annual fuel expenditures that account for nearly a quarter of cargo companies' annual budgets. Most technology that has aimed to do this has primarily been focused on either improvements in engine efficiency or reduction of aerodynamic drag. This thesis addresses a somewhat different approach: the optimization of vehicle dynamics in order to realize fuel savings. </div><div><br></div><div>Through partnerships with Peloton Technology and Cummins, tests and simulations were conducted on corridors with grades up to 5% that indicate fuel savings of up to 14.4% can be achieved through the combination of three strategies: two-truck platooning, long-horizon predictive cruise control (LHPCC), and simultaneous shifting. Two-truck platooning is the act of drafting a rear truck behind a front truck. It has been shown that this not only reduces the drag of the follow vehicle, but also that of the lead vehicle. LHPCC is an optimization of the lead truck's velocity over a given corridor to get "from point A to point B" in the most efficient way possible whilst doing so with a trip time constraint. Last is the use of simultaneous shifting, which allows the follow vehicle to maintain the proper platoon gap distance behind</div><div>the lead truck.</div>

Robust and timely perception of the environment is an essential requirement of all autonomous and semi-autonomous systems. This necessity has been the main factor behind the rapid growth and adoption of LiDAR sensors within the ADAS sensor suite. In this thesis, we develop a fast and accurate 3D object detector that converts raw point clouds collected by LiDARs into sparse occupancy cuboids to detect cars and other road users using deep convolutional neural networks. The proposed pipeline reduces the runtime of PointPillars by 43% and performs on par with other state-of-the-art models. We do not gain improvements in speed by compromising the network's complexity and learning capacity but rather through the use of an efficient input encoding procedure. In addition to rigorous profiling on three different platforms, we conduct a comprehensive error analysis and recognize principal sources of error among the predicted attributes. Even though point clouds adequately capture the 3D structure of the physical world, they lack the rich texture information present in color images. In light of this, we explore the possibility of fusing the two modalities with the intent of improving detection accuracy. We present a late fusion strategy that merges the classification head of our LiDAR-based object detector with semantic segmentation maps inferred from images. Extensive experiments on the KITTI 3D object detection benchmark demonstrate the validity of the proposed fusion scheme.

No description available.

No description available.

Self-driving is a technology that has been envisioned in science fiction movies or in speculative design for quite some time. However, it is one of the few future technologies that is relatively easy to imagine, but very difficult to implement it into reality due to complexity coming from variability in AI. This discrepancy between reality and imagination is what makes achieving trust in self-driving vehicles more challenging, especially regarding the fact that driving is regarded as a daily task for some people. Keeping into consideration how most of the other projects done to enhance trust in automation deals with full automation, this thesis focuses how trust can be defined in semi autonomous vehicles. This middle ground setting with humans and AI systems working together needs more factors to be considered to make it autonomous, at the same time requiring a higher level of trust from drivers. An additional layer of a takeover situation from driver to AI and vice versa in a semi-autonomous setting would require more level of trust than a full self-driving vehicle where drivers do not have to control anything.Volvo Cars, an automobile manufacturer brand that has its strong focus on safety, was collaborated with in this project to support developing a notion of trust in autonomous systems. The purpose of this collaboration with Volvo Cars was to receive support in any expert knowledge in the mobility field and to create a project that is relevant to the current development state and future vision of autonomous vehicles. In order to provide an environment where drivers can calibrate trust inside vehicles, FiDO, a tangible driving assistant for building trust, was designed through a participatory design process. FiDO provides an environment for setting mutual expectation between driver and vehicle through communicating vehicle’s status and driver’s feedback with poetic visuals. FiDO learns from driver’s behaviors and their direct feedback, which provides personalized content and autonomous driving as an outcome of learning. FiDO’s usage can be adjusted based on driver’s trust level and characteristics of the service of where automation technology is used.This thesis does not cover the entire notion of trust in automation, but focuses particularly on building trust from a driver’s point of view. With including users throughout the process, this is a proof of concept how automation technology and notion of trust can be built with driver’s participation. Although detailed technological feasibility of including both humans and AI in one place to build an autonomous system were not considered into practical levels, this thesis emphasizes how we can also establish trust voluntarily from a user’s point of view.

No description available.

No description available.

The main research objective of this dissertation is to understand the sensing and communication challenges to achieving cooperative perception among autonomous vehicles, and then, using the insights gained, guide the design of the suitable format of data to be exchanged, reliable and efficient data fusion algorithms on vehicles. By understanding what and how data are exchanged among autonomous vehicles, from a machine learning perspective, it is possible to realize precise cooperative perception on autonomous vehicles, enabling massive amounts of sensor information to be shared amongst vehicles. I first discuss the trustworthy perception information sharing on connected and autonomous vehicles. Then how to achieve effective cooperative perception on autonomous vehicles via exchanging feature maps among vehicles is discussed in the following. In the last methodology part, I propose a set of mechanisms to improve the solution proposed before, i.e., reducing the amount of data transmitted in the network to achieve an efficient cooperative perception. The effectiveness and efficiency of our mechanism is analyzed and discussed.

Long range path planning for an autonomous ground vehicle with minimal a-priori data is still very much an open problem. Previous research has demonstrated that least cost paths generated from aerial LIDAR and GIS data could play a role in automatically determining suitable routes over otherwise unknown terrain. However, most of this research has been theoretical. Consequently, there is very little literature the effectiveness of these techniques in plotting paths of an actual autonomous vehicle. This research aims to develop an algorithm for using aerial LIDAR and imagery to plan paths for a full size autonomous car. Methods of identifying obstacles and potential roadways from the aerial LIDAR and imagery are reviewed. A scheme for integrating the path planning algorithms into the autonomous vehicle existing systems was developed and eight paths were generated and driven by an autonomous vehicle. The paths were then analyzed for their drivability and the model itself was validated against the vehicle measurements. The methods described were found to be suitable for generating paths both on and off road. / Master of Science

This research study presents two control solutions, PID and the novel hyper-clothoid control strategy, to autonomously navigate a car. These waypoint navigation solutions smoothly connect the given waypoints with C1 continuity using Hermite cubic splines which is used as a reference path for the controller to track. The PID controller uses lateral and heading error to generate a steering profile for the vehicle to track the constructed reference path. A novel real time solution is presented as the second control strategy which involves constructing clothoids to generate a steering profile. The resultant car trajectory preserves curvature and curvature rate continuity. A simulation test bench was developed in MATLAB and Simulink. Six benchmark waypoint datasets have been used for regression testing and validating the algorithms. Both the proposed control strategies have been implemented on a 2017 GM Chevy Bolt EV. A real time operating system (QNX) has been used and was time-synced with the localization suite in the test vehicle. Closed loop results with accuracies up to 50 cm of lateral error have been achieved using the test vehicle. / Doctor of Philosophy / The research into self-driving cars has been one of the most sought out areas these past couple of decades. There are many components into building a self-driving car - Sensing, Perception, Localization, Navigation. Lot of strategies have been developed over the years with waypoint navigation being the most widely used for navigation an autonomous vehicle. Waypoint Navigation utilizes GPS data to move the car from one point to the other. The traditional process of this strategy involves two parts - curve fitting between waypoints and using a control scheme to track the path with the car. Numerous methods have been developed to fit a curve in between two points. Most of these methods use a variant of 3rd degree or higher order polynomials . Also different control strategies have been developed to track the generated path. Model predictive control strategies are among the popular control architectures used for this purpose. This work proposes a novel method to track a path using clothoids. The proposed algorithm has a novel approach of integrating the path construction and control strategy. The algorithm also has a low computational requirement making it highly suitable for implementation in real-time.