Localization is one of the most important capabilities of mobile robots. Thanks to the fast development of embedded computing hardware in recent years, many localization solutions, such as simultaneous localization and mapping (SLAM), have been vastly investigated. However, popular localization solutions rely heavily on the working environment and are not applicable to scenarios such as search and rescue in the wild, where the working environment is not accessible before the localization operation or where the environment lacks information on features and textures. The thesis thus proposes a design for an innovative localization sensor and a collaborative pose estimation scheme using the localization sensor in order to alleviate the reliance on information from the environment, while providing reliable and accurate pose estimates for mobile robots. The proposed collaborative pose estimation scheme is comprised of individual and collaborative landmark position estimation, localization sensor inter-calibration, and collaborative sensor pose estimation, among which the inter-calibration process ensures that the sensor provides capability to also estimate orientations. In the collaborative scheme, multiple instances of the proposed sensor collaborate to estimate their respective poses by measuring the relative distance and angle among them, where the measurement errors are characterized as Gaussian white noise. Two instances of the proposed localization sensor are implemented, and the collaborative scheme is tested with the instances in the thesis. Both sensor instances reliably and accurately estimate the position of a stationary landmark, and it is demonstrated that the collaboratively estimated position estimate is more accurate than its individual counterpart. Additionally, the two instances also demonstrate their ability to track and estimate the position of a moving landmark. Lastly, the inter-calibration is experimentally validated with the instances with satisfactory performance. The experimental results presented in this work confirm the feasibility and usability of the proposed collaborative pose estimation scheme in a wide range of robotic applications.

Robots have become increasingly adept at performing a wide variety of tasks in the world. However, many of these tasks can benefit tremendously from having more than a single robot simultaneously working on the problem. Multiple robots can aid in a search and rescue mission each scouting a subsection of the entire area in order to cover it quicker than a single robot can. Alternatively, robots with different abilities can collaborate in order to achieve goals that individually would be more difficult, if not impossible, to achieve. In these cases, multi-robot collaboration can provide benefits in terms of shortening search times, providing a larger mix of sensing, computing, and manipulation capabilities, or providing redundancy to the system for communications or mission accomplishment. One principle drawback of multi-robot systems is how to efficiently and effectively generate plans that use each of the team members to their fullest extent, particularly with a heterogeneous mix of capabilities. Towards this goal, I have developed a series of planning algorithms that incorporate this collaboration into the planning process. Starting with systems that use collaboration in an exploration task I show teams of homogeneous ground robots planning to efficiently explore an initially unknown space. These robots share map information and in a centralized fashion determine the best goal location for each taking into account the information gained by other robots as they move. This work is followed up with a similar exploration scheme but this time expanded to a heterogeneous air-ground robot team operating in a full 3-dimensional environment. The extra dimension adds the requirement for the robots to reason about what portions of the environment they can sense during the planning process. With an air-ground team, there are portions of the environment that can only be sensed by one of the two robots and that information informs the algorithm during the planning process. Finally, I extend the air-ground robot team to moving beyond merely collaboratively constructing the map to actually using the other robots to provide pose information for the sensor and computationally limited team members. By explicitly reasoning about when and where the robots must collaborate during the planning process, this approach can generate trajectories that are not feasible to execute if planning occurred on an individual robot basis. An additional contribution of this thesis is the development of the State Lattice Planning with Controller-based Motion Primitives (SLC) framework. While SLC was developed to support the collaborative localization of multiple robots, it can also be used by a single robot to provide a more robust means of planning. For example, using the SLC algorithm to plan using a combination of vision-based and metric-based motion primitives allows a robot to traverse a GPS-denied region.

High-accuracy localization is vital for many applications and is a fundamental prerequisite for enabling autonomous missions. Modern navigation systems often rely heavily on Global Navigation Satellite Systems (GNSS) for achieving high localization accuracy over extended periods of time, which has necessitated alternative localization methods that can be used in GNSS-disturbed environments. One popular alternative that has emerged is Collaborative Localization (CL), which is a method where agents of a swarm combine knowledge of their own state with relative measurements of other agents to achieve a localization accuracy that is better than what a single agent can achieve on its own. Performing this in a decentralized manner introduces the challenge of how to account for unknown inter-agent correlations, which typically leads to the need for using conservative fusion methods such as Covariance Intersection (CI) to preserve consistency. Many existing CL algorithms that utilize CI assume agents to have perception systems capable of identifying the relative position of other swarm members. These algorithms do therefore not work in systems where, e.g., agents are only capable of measuring range to each other. Other CI algorithms that support more generic measurement models can require large amounts of data to be exchanged when agents communicate, which could lead to issues in bandwidth-limited systems. This thesis develops a consistent decentralized collaborative localization algorithm based on CI that supports range-only measurements between agents and requires a communication effort that is constant in the number of agents in the swarm. The algorithm, referred to as the PSCI algorithm, was found to maintain satisfactory performance in various scenarios but exhibits slightly increased sensitivity to the measurement geometry compared to an already existing, more communication-heavy, CI-based algorithm. Moreover, the thesis highlights the impact of linearization errors in range-only CL systems and shows that performing CI-fusion before the range-observation measurement update, with a clever choice of CI cost function, can reduce linearization errors for the PSCI algorithm. A comparison between the PSCI algorithm and an already existing algorithm, referred to as the Cross-Covariance Approximation (CCA) algorithm, has further been conducted through a sensitivity analysis with respect to communication rate and the number of GNSS agents. The simulation results indicate that the PSCI algorithm exhibits diminishing improvement in Root Mean Square Error (RMSE) with increased communication rates, while the RMSE of the CCA algorithm reaches a local minimum, subsequently showing overconfidence with higher rates. Lastly, evaluation under a varying number of GNSS agents indicates that cooperative benefits for the PSCI filter are marginal when uncertainty levels are uniform across agents. However, the PSCI algorithm demonstrates superior performance improvements with an increased number of GNSS agents compared to the CCA algorithm, attributed to the overconfidence of the latter.

In this thesis two collaborative localization techniques are studied: multidimensional scaling (MDS) and maximum likelihood estimator (MLE). A synthesis of a new location estimation method through a serial integration of these two techniques, such that an estimate is first obtained using MDS and then MLE is employed to fine-tune the MDS solution, was the subject of this research using various simulation and experimental studies. In the simulations, important issues including the effects of sensor node density, reference node density and different deployment strategies of reference nodes were addressed. In the experimental study, the path loss model of indoor environments is developed by determining the environment-specific parameters from the experimental measurement data. Then, the empirical path loss model is employed in the analysis and simulation study of the performance of collaborative localization techniques.