Robots in human spaces such as malls, airports, and stadiums are already deployed in the real world to perform a variety of tasks, such as guidance, or cleaning. However, these deployed systems often come with stipulations of high cost sensing, re-engineering the space itself to accommodate the robot, providing costly a priori information, and/or expert intervention. I advance towards accessible, human-centered robot systems using theory and tools that use context to avoid these limitations. I use insights on how humans interact with their spaces, incorporate human perspectives through user studies to motivate the approach and to evaluate the system. I aim for real-time execution with limited on-board computing power by centering my system around a theme of minimalism, where, at each stage of system development, the robot should use the bare minimum amount of sensing and information needed to achieve its goals. This thesis primarily discusses the development of a robot that navigates in public spaces without a map, using only signs designed for humans and relying solely on local sensing. The system is developed from proofs to real perception and is evaluated in real transit stations by real people. We also present a field study that sheds light on how humans naturally communicate with robots about the space they share, and how insights from the study align with design choices for our robot system. As a result, the intended system can be dropped in completely unseen spaces and perform efficient navigation out-of-the-box. This thesis also includes descriptions and references for the data from each stage of system development and additional work beyond the airport robot system that aligns with the themes of human context and minimalism.