Timezone: »

Constructing Skill Trees for Reinforcement Learning Agents from Demonstration Trajectories
George Konidaris · Scott R Kuindersma · Andrew G Barto · Roderic A Grupen

Tue Dec 07 12:00 AM -- 12:00 AM (PST) @

We introduce CST, an algorithm for constructing skill trees from demonstration trajectories in continuous reinforcement learning domains. CST uses a changepoint detection method to segment each trajectory into a skill chain by detecting a change of appropriate abstraction, or that a segment is too complex to model as a single skill. The skill chains from each trajectory are then merged to form a skill tree. We demonstrate that CST constructs an appropriate skill tree that can be further refined through learning in a challenging continuous domain, and that it can be used to segment demonstration trajectories on a mobile manipulator into chains of skills where each skill is assigned an appropriate abstraction.

Author Information

George Konidaris (Brown University)
Scott R Kuindersma (University of Massachusetts Amherst)
Andrew G Barto (University of Massachusetts)
Roderic A Grupen (UMass Amherst)

More from the Same Authors