Constructing a system of intelligent robotic mapping agents that can function in an unstructured and unknown environment is a challenging task. With the exploration of our solar system as well as our own planet requiring more robust mapping agents, and with the drastic drop in the price of technology versus the gains in performance, robotic mapping is becoming a focus of research like never before. Efforts are underway to send mobile robots to map bodies within our solar system. While much of the research in robotic map construction has been focused on building maps used by the robotic agents themselves, very little has been done in building maps usable by humans. And yet it is the human that drives the need for mapping solutions.; We propose a computational framework for building mobile robotic mapping systems to be deployed in unknown environments. This is the first work known to address the general problem of mapping in unknown terrain under the affect of error in readings, operations and systems that employs more than a single robot. The system draws upon the strengths from research in various robotic related areas by selecting those components and ideas that show promise when applied to mapping for human reading via a distributed network of heterogeneous mobile robots. This application of multiple mobile robots and the application to human end-users is a new direction in robotics research. We also propose and develop a new paradigm for storing mapping-agent generated data in a way that allows rapid map construction and correction to compensate for detected errors. We experimentally test the paradigm on a simulated robotic environment and analyze the results and show that there is a definite gain from correction, particularly in error rich environments. We also develop methods by which to apply corrections to the map and test their effectiveness. Finally we propose some extensions to this work and suggest research in areas not completely covered by our discussion.
展开▼
