There has been an increasing interest in visual sensors and vision-based solutions for single and multi-robot systems. Vision-based sensors, e.g., traditional RGB cameras, grant rich semantic information and accurate directional measurements at a relatively low cost; however, such sensors have two major drawbacks. They do not generally provide reliable depth estimates, and typically have a limited field of view. These limitations considerably increase the complexity of controlling multiagent systems. This thesis studies some of the underlying problems in vision-based multiagent control and mapping. The first contribution of this thesis is a method for restoring bearing rigidity in non-rigid networks of robots. We introduce means to determine which bearing measurements can improve bearing rigidity in non-rigid graphs and provide a greedy algorithm that restores rigidity in 2D with a minimum number of added edges. The focus of the second part is on the formation control problem using only bearing measurements. We address the control problem for consensus and formation control through non-smooth Lyapunov functions and differential inclusion. We provide a stability analysis for undirected graphs and investigate the derived controllers for directed graphs. We also introduce a newer notion of bearing persistence for pure bearing-based control in directed graphs. The third part is concerned with the bearing-only visual homing problem with a limited field of view sensor. In essence, this problem is a special case of the formation control problem where there is a single moving agent with fixed neighbors. We introduce a navigational vector field composed of two orthogonal vector fields that converges to the goal position and does not violate the field of view constraints. Our method does not require the landmarks' locations and is robust to the landmarks' tracking loss. The last part of this dissertation considers outlier detection in pose graphs for Structure from Motion (SfM) and Simultaneous Localization and Mapping (SLAM) problems. We propose a method for detecting incorrect orientation measurements before pose graph optimization by checking their geometric consistency in cycles. We use Expectation-Maximization to fine-tune the noise's distribution parameters and propose a new approximate graph inference procedure specifically designed to take advantage of evidence on cycles with better performance than standard approaches. These works will help enable multi-robot systems to overcome visual sensors' limitations in collaborative tasks such as navigation and mapping.

<p>Alternate power sources in automotive class-8 trucking industry is a major focus of research in recent days. Green house gasses, oxides of Nitrogen(NOx), Oxides of Sulphur(SOx), hydrocarbons and particulate matter are major concerns contributing to the shift in alternate fuel strategies. Another direct relation to move to an alternate power strategy is the reduction in net fuel consumption which in turn implicitly improves the emission components.</p> <p>A holistic approach is needed while designing a modern class-8 vehicle. A variety of system architecture, control algorithms, diagnostic levers are needed to be manipulated to achieve the best of blends amongst Total Cost of Ownership (TCO), Drivability, Fuel</p> <p>Economy, Emissions Compliant, Hauling Capacity, etc. The control and system levers are not mutually exclusive and there is a strong correlation amongst all these control and system components. In order to achieve a consensus amongst all these levers to achieve a common objective, is a challenging and complex problem to solve. It is often required to shift the algorithm strategy to predictive information based rather than reactive logic. Predictively modulating and manipulating control logic can help with better fuel efficient solution along with emissions improvement. A further addition to the above challenge is when we add a fleet of vehicle to the problem. So, the problem now is to optimize a control action for a fleet</p> <p>of vehicles and design/select the correct component size. A lot of research has been done and is still underway to use a 48V hybrid system with a small battery using a simple charge sustaining SOC control strategy. This will make the system light enough not to compromise on the freight carrying capacity as well as give some extra boost during the high torque requirement sections in the route for a better fuel and emissions efficient solution. In this work a P2 type 48V hybrid system is used which is side mounted to the transmission via a gear system. The selection of the system and components enables the usage of different control strategies such as neutral coasting and Engine off coasting. This architecture with a traditional 12-15L Internal combustion engine along with a mild 48V hybrid system provides the most viable selection for a long haul class-8 application and is used in this work. It is also possible to identify other component sizes along with architectures for new configurations. The framework in this research work can help develop the study for different component sizing. While this research work is focused towards building a framework for achieving predictive control in a 3 truck platooning system using multi-agent based control, the other supporting work done also helps understand the optimal behavior of the interacting multiple controls when the corridor information such as road grade and route speed limit are known a-priori, in a single vehicle. The build up of this work analyzes an offline simulation of a 4 control optimal solution for a single hybrid truck and then extend the optimal controls to a 3 truck platoon. In the single truck, this research will help identify the interacting zones in the route where the various control actions will provide the best cost benefits which is fuel economy. These benefits are associated as a function of exogenous look ahead information such as grade and speed limit. Further it is also possible to identify the optimal behavior and the look ahead horizon required for achieving that. In other words the optimal behavior and benefits associated with the global solution can be accomplished by implementing rule based control system with a look ahead horizon of 2-5 km. If this would not have been the case then it is almost impossible to design a predictive controller based on the entire route information which can stretch up to hundreds of kilometers. Optimal algorithms of such prediction horizon are not feasible to be implemented in real time controllers. This research work will also help understand the interaction between different active control actions such as predictive speed modulation, gear shift, coasting and power split with passive control levers such as slow down due to hybrid regeneration, hybrid boost during coasting, etc. This will help in architecting a system involving component specifications, active optimal control, look ahead information, hybrid system strength, etc, working in close interaction with each other. Though we analyze these predictive behavior for a single vehicle as a supporting work the prime objective is to include these predictive levers in a platooning system using an agent based method. This multi-agent based technique will help analyze the behavior of multiple trucks in a platoon in terms of fuel efficient safe operation. The focus of this research work is to not directly come up with a controller or strategy but rather to understand the optimality of this control levers for a multi-vehicle platoon system given a look ahead information is available. The research shows that predictive information will help in gaining fuel economy for a platoon of class-8 mild hybrid trucks. It also highlights the challenges in doing so and what needs to be traded off in order to achieve the net fuel benefit.</p>