Short answers, pulled from the story.
Pieter Abbeel and Andrew Ng introduced the concept of apprenticeship learning at Stanford University and the University of California, Berkeley. They defined this process as a form of supervised learning where the training dataset consists of task executions by a demonstration teacher.
Stuart J. Russell proposed using inverse reinforcement learning to derive reward functions from observed human behavior instead of using rewards to shape behavior. Robots observe people to figure out what goals their actions seem to be trying to achieve based on behavioral measurements over time.
Pieter Abbeel, Adam Coates, and Andrew Ng applied apprenticeship learning to autonomous helicopter control in 2010. They published their findings in the International Journal of Robotics Research volume 29 issue 13 after teaching helicopters to perform complex aerial maneuvers like in-place flips and loops.
Adrian Stoica published his PhD thesis in 1995 focusing on anthropomorphic robots learning by imitation. His work represented one of the first attempts to teach humanoid systems generalized plans from limited examples as demonstrated in a 1994 experiment showing a humanoid learning a repetitive ball collection task from only two human demonstrations.
Stefan Schaal worked with the Sarcos robot-arm in 1997 to solve the pendulum swingup task by recording human movements over three seconds at the y-axis. This data produced a pattern showing angle changes relative to time intervals ranging from zero to three seconds.