Humans are at the center of robot learning. Whether providing training data, initializing environments, or tuning robots, humans are critical for effective learning. But in research we often find disconnects between how robots learn and how humans interact. From the human's perspective, users expect robots to understand their actions and robustly perform tasks --- i.e., people think robots have abilities that match our "common sense.'' From the robot's perspective, learners expect humans to provide large amounts of near-optimal data --- but these detailed, intricate instructions can be difficult or unnatural for non-experts. 

Our proposed workshop will reunite HRI with robot learning by bringing together diverse experts from both perspectives. In human-robot interaction, we will cover best practices for user studies and explore how to design learners that are intuitive for everyday users. In robot learning, we will summarize cutting-edge frameworks, and assess what people should realistically expect from these systems. Finally, at the intersection of both fields, we will explore how human teachers and robot learners can mutually improve one another's performance. Our objectives include i) understanding how to combine technical learning advances with human-centered studies and ii) a position paper that documents trends, challenges, and interdisciplinary directions for future research.