Titlebook: Robot Learning from Human Teachers; Sonia Chernova,Andrea L. Thomaz Book 2014 Springer Nature Switzerland AG 2014

Graduated

线

Introduction, the real world. Today, and for the foreseeable future, it is not possible to go to a store and bring home a robot that will clean your house, cook your breakfast, and do your laundry. These everyday tasks, while seemingly simple, contain many variations and complexities that pose insurmountable challenges for today’s machine learning algorithms.

exostosis

Limpid

Learning Low-Level Motion Trajectories, . in which they would be used (covered in Chapter 5). In the literature there are several different names given to this class of “low-level” action learning, thus in this chapter we use the terms . and . interchangeably.

GIBE

Learning High-Level Tasks,ning a reactive task policy representing a functional mapping of states to actions, learning a task plan, and learning the task objectives. We go on to discuss the role that feature selection, reference frame identification and object affordances play in the learning process.

Distribution

主讲人

Human Social Learning, process. Although robots can also learn from observing demonstrations not directed at them, albeit less efficiently, the scenario we address here is primarily the one where a person is explicitly trying to teach the robot something in particular.

affect

你敢命令

Learning Low-Level Motion Trajectories,algorithm can be designed to work with. We now turn our attention to the wide range of algorithms for building skill and task models from demonstration data. In this chapter we focus on approaches that learn new motions or primitive actions. The motivation behind learning new motions is typically th

Systemic

Learning High-Level Tasks, (Figure 5.1). While the line between high-level and low-level learning is not concrete, the distinction we make here is that techniques in this chapter assume the existence of a discrete set of action primitives that can be combined to perform a more complex behavior. As in the previous chapter, we