Skip to Main content Skip to Navigation
Conference papers

Modulated Policy Hierarchies

Abstract : Solving tasks with sparse rewards is a main challenge in reinforcement learning. While hierarchical controllers are an intuitive approach to this problem, current methods often require manual reward shaping, alternating training phases, or manually defined sub tasks. We introduce modulated policy hierarchies (MPH), that can learn end-to-end to solve tasks from sparse rewards. To achieve this, we study different modulation signals and exploration for hierarchical controllers. Specifically, we find that communicating via bit-vectors is more efficient than selecting one out of multiple skills, as it enables mixing between them. To facilitate exploration, MPH uses its different time scales for temporally extended intrinsic motivation at each level of the hierarchy. We evaluate MPH on the robotics tasks of pushing and sparse block stacking, where it outperforms recent baselines.
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-01963580
Contributor : Alexander Pashevich <>
Submitted on : Friday, December 21, 2018 - 1:53:44 PM
Last modification on : Friday, April 17, 2020 - 11:16:02 AM

Links full text

Identifiers

  • HAL Id : hal-01963580, version 1
  • ARXIV : 1812.00025

Collections

Citation

Alexander Pashevich, Danijar Hafner, James Davidson, Rahul Sukthankar, Cordelia Schmid. Modulated Policy Hierarchies. Deep Reinforcement Learning Workshop at NeurIPS 2018, Dec 2018, Montreal, Canada. ⟨hal-01963580⟩

Share

Metrics

Record views

235