English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

We propose a novel approach to learn goal-conditioned policies for locomotionin a batch RL setting. The batch data is collected by a policy that is notgoal-conditioned. For the locomotion task, this translates to data collectionusing a policy learnt by the agent for walking straight in one direction, andusing that data to learn a goal-conditioned policy that enables the agent towalk in any direction. The data collection policy used should be invariant tothe direction the agent is facing i.e. regardless of its initial orientation,the agent should take the same actions to walk forward. We exploit thisproperty to learn a goal-conditioned policy using two key ideas: (1) augmentingdata by generating trajectories with the same actions in different directions,and (2) learning an encoder that enforces invariance between these rotatedtrajectories with a Siamese framework. We show that our approach outperformsexisting RL algorithms on 3-D locomotion agents like Ant, Humanoid andMinitaur.

Goal-conditioned Batch Reinforcement Learning for Rotation Invariant Locomotion