Robots are a unique form of technology that hold the potential to significantly affect society as they become more widespread in everyday life. Currently, they are in use in domains including healthcare, rehabilitation, assisted living, entertainment, education, and homes. Within these domains, robots interact with people from all kinds of backgrounds. Each person has a large set of qualities that make them unique. These qualities affect both how people perceive robots, and also how robots perceive people. Furthermore, the environmental context in which robots and people are collaborating can also play a role in this mutual behavioral understanding. Thus, to ensure robots are both functional and well-accepted, roboticists should consider taking these factors into account. The motivation of my research is to design adaptable, intelligent, social robots, able to sense and respond to people contingently and appropriately. This will enable more intuitive interaction between robots and people. I focus specifically on identifying human behavioral metrics which future robots could one day be trained to identify as a means for enabling this understanding. This dissertation outlines four human-robot interaction (HRI) experiments which explore human behavioral metrics. The first experiment highlights differences in how people perceive gestures and speech by a humanoid robot actor compared to a human actor. The second experiment centers on a robot that initiates interactions with people, with the objective of collecting naturalistic data to determine social context. The third experiment focuses on how personality traits may affect patience when teaching an autonomous, mistake-prone robot. The last experiment explores how human teachers respond to both correct and incorrect robot actions to eventually allow robots to automatically detect when they have made a mistake. My work has multiple contributions for HRI. First, I identify key differences between peoples perceptions of robot and human behavior, which are important to consider when programming robot communicative expressions. Second, I explore how multiple dimensions of human personality affect HRI, which differs from previous HRI work that focused on only one or two dimensions at a time. During the course of my work, I built autonomous robots that robustly responded to people in real-time, as opposed to the majority of HRI research that involves operator-control robots. My third contribution is to describe these systems, and identify the advantages and challenges inherent in this kind of research. Fourth, I discover ways that robots can enable human teachers by providing feedback to assist in the learning process, which is is a necessary step to achieve natural, efficient, and fluid interactions with adaptable robots. I also designed two experimental testbeds which focus on peoples responses to robot mistakes, which help enable future research in this area. The analysis of observable human actions will enable the creation of human behavioral metrics for HRI that can be incorporated into future robot algorithms. This work will inform the design of personalized robots in the future, which can both reflect the individuality of their users, implicitly learn from their mistakes, and transition into machines that people will want to interact with.