Master of Science
Imagine, if we want to explore into an unknown and unexplored environment, the first thing we will need is a map of the environment, hence with the help of local/global positioning system and mapping technology, a robot can map the environment that can further utilized for exploration. Now another question that can arise here is what if local/global positioning system is not available, while we have the map of the environment, hence for a robot to know its exact location at any instance of time into that environment its needs to localize itself. But what if we have neither of the two things maps as well as access to local/global positioning system, in such situations we need to map as well localize our robot into that unknown and unexplored environment at the same time and hence the concept of Simultaneous Localization And Mapping (SLAM) came into existence.
SLAM consists in the simultaneous construction of a map of the
environment, and the estimation of the state (localization) of the robot moving within it. The SLAM community in robotics has made some astonishing progress over the past 30 years, enabling large-scale real-world applications, and witnessing a steady transition of this technology to the robotics industry. In this thesis we study a novel approach to construct a metric embedding of landmarks observed by a robot equipped with a noisy range sensor, navigating in an unknown environment. In every iteration of the algorithm, a graph constituting of the landmarks and observation points as vertices are modeled as a spring-mass-damper system and its dynamic simulation is performed to obtain the optimal lengths of the links in the graph. The dynamic simulation is run every time a new set of observations are obtained, with the result of the previous dynamic simulation used as its initial condition. This incrementally constructs an approximate metric representation of the landmarks and robot poses, in effect giving us a metric map of the landmarks and allowing us to localize the robot relative to that.
Jagina, Sagar Kalpesh, "Landmark-based Mapping, Localization, and Navigation via Dynamic Simulation" (2019). Theses and Dissertations. 5669.