The Intentional Unintentional Agent: Learning to Solve Many Continuous Control Tasks Simultaneously

Serkan Cabi, Sergio Gómez Colmenarejo, Matthew W. Hoffman, Misha Denil, Ziyu Wang, Nando Freitas
Proceedings of the 1st Annual Conference on Robot Learning, PMLR 78:207-216, 2017.

Abstract

This paper introduces the Intentional Unintentional (IU) agent. This agent endows the deep deterministic policy gradients (DDPG) agent for continuous control with the ability to solve several tasks simultaneously. Learning to solve many tasks simultaneously has been a long-standing, core goal of artificial intelligence, inspired by infant development and motivated by the desire to build flexible robot manipulators capable of many diverse behaviours. We show that the IU agent not only learns to solve many tasks simultaneously but it also learns faster than agents that target a single task at-a-time. In some cases, where the single task DDPG method completely fails, the IU agent successfully solves the task. To demonstrate this, we build a playroom environment using the MuJoCo physics engine, and introduce a grounded formal language to automatically generate tasks.

Cite this Paper


BibTeX
@InProceedings{pmlr-v78-cabi17a, title = {The Intentional Unintentional Agent: Learning to Solve Many Continuous Control Tasks Simultaneously}, author = {Cabi, Serkan and Colmenarejo, Sergio Gómez and Hoffman, Matthew W. and Denil, Misha and Wang, Ziyu and Freitas, Nando}, booktitle = {Proceedings of the 1st Annual Conference on Robot Learning}, pages = {207--216}, year = {2017}, editor = {Levine, Sergey and Vanhoucke, Vincent and Goldberg, Ken}, volume = {78}, series = {Proceedings of Machine Learning Research}, month = {13--15 Nov}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v78/cabi17a/cabi17a.pdf}, url = {https://proceedings.mlr.press/v78/cabi17a.html}, abstract = {This paper introduces the Intentional Unintentional (IU) agent. This agent endows the deep deterministic policy gradients (DDPG) agent for continuous control with the ability to solve several tasks simultaneously. Learning to solve many tasks simultaneously has been a long-standing, core goal of artificial intelligence, inspired by infant development and motivated by the desire to build flexible robot manipulators capable of many diverse behaviours. We show that the IU agent not only learns to solve many tasks simultaneously but it also learns faster than agents that target a single task at-a-time. In some cases, where the single task DDPG method completely fails, the IU agent successfully solves the task. To demonstrate this, we build a playroom environment using the MuJoCo physics engine, and introduce a grounded formal language to automatically generate tasks.} }
Endnote
%0 Conference Paper %T The Intentional Unintentional Agent: Learning to Solve Many Continuous Control Tasks Simultaneously %A Serkan Cabi %A Sergio Gómez Colmenarejo %A Matthew W. Hoffman %A Misha Denil %A Ziyu Wang %A Nando Freitas %B Proceedings of the 1st Annual Conference on Robot Learning %C Proceedings of Machine Learning Research %D 2017 %E Sergey Levine %E Vincent Vanhoucke %E Ken Goldberg %F pmlr-v78-cabi17a %I PMLR %P 207--216 %U https://proceedings.mlr.press/v78/cabi17a.html %V 78 %X This paper introduces the Intentional Unintentional (IU) agent. This agent endows the deep deterministic policy gradients (DDPG) agent for continuous control with the ability to solve several tasks simultaneously. Learning to solve many tasks simultaneously has been a long-standing, core goal of artificial intelligence, inspired by infant development and motivated by the desire to build flexible robot manipulators capable of many diverse behaviours. We show that the IU agent not only learns to solve many tasks simultaneously but it also learns faster than agents that target a single task at-a-time. In some cases, where the single task DDPG method completely fails, the IU agent successfully solves the task. To demonstrate this, we build a playroom environment using the MuJoCo physics engine, and introduce a grounded formal language to automatically generate tasks.
APA
Cabi, S., Colmenarejo, S.G., Hoffman, M.W., Denil, M., Wang, Z. & Freitas, N.. (2017). The Intentional Unintentional Agent: Learning to Solve Many Continuous Control Tasks Simultaneously. Proceedings of the 1st Annual Conference on Robot Learning, in Proceedings of Machine Learning Research 78:207-216 Available from https://proceedings.mlr.press/v78/cabi17a.html.

Related Material