Recent advances in computing and robotics offer significant potential for improved autonomy in the operation and utilization of today's buildings. Examples of such building environment functions that could be improved through automation include: a) building performance monitoring for real-time system control and long-term asset management; and b) assisted indoor navigation for improved accessibility and wayfinding. To enable such autonomy, algorithms related to task allocation, path planning, and navigation are required as fundamental technical capabilities. Existing algorithms in these domains have primarily been developed for outdoor environments. However, key technical challenges that prevent the adoption of such algorithms to indoor environments include: a) the inability of the widely adopted outdoor positioning method (Global Positioning System - GPS) to work indoors; and b) the incompleteness of graph networks formed based on indoor environments due to physical access constraints not encountered outdoors.;The objective of this dissertation is to develop general and scalable task allocation, path planning, and navigation algorithms for indoor mobile co-robots that are immune to the aforementioned challenges. The primary contributions of this research are: a) route planning and task allocation algorithms for centrally-located mobile co-robots charged with spatiotemporal tasks in arbitrary built environments; b) path planning algorithms that take preferential and pragmatic constraints (e.g., wheelchair ramps) into consideration to determine optimal accessible paths in building environments; and c) navigation and drift correction algorithms for autonomous mobile robotic data collection in buildings.;The developed methods and the resulting computational framework have been validated through several simulated experiments and physical deployments in real building environments. Specifically, a scenario analysis is conducted to compare the performance of existing outdoor methods with the developed approach for indoor multi-robotic task allocation and route planning. A simulated case study is performed along with a pilot experiment in an indoor built environment to test the efficiency of the path planning algorithm and the performance of the assisted navigation interface developed considering people with physical disabilities (i.e., wheelchair users) as building occupants and visitors. Furthermore, a case study is performed to demonstrate the informed retrofit decision-making process with the help of data collected by an intelligent multi-sensor fused robot that is subsequently used in an EnergyPlus simulation. The results demonstrate the feasibility of the proposed methods in a range of applications involving constraints on both the environment (e.g., path obstructions) and robot capabilities (e.g., maximum travel distance on a single charge). By focusing on the technical capabilities required for safe and efficient indoor robot operation, this dissertation contributes to the fundamental science that will make mobile co-robots ubiquitous in building environments in the near future.
展开▼
