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

Vanilla Reinforcement Learning (RL) can efficiently solve complex tasks butdoes not provide any guarantees on system behavior. To bridge this gap, wepropose a three-step safe RL procedure for continuous action spaces thatprovides probabilistic guarantees with respect to temporal logicspecifications. First, our approach probabilistically verifies a candidatecontroller with respect to a temporal logic specification while randomizing thecontrol inputs to the system within a bounded set. Second, we improve theperformance of this probabilistically verified controller by adding an RL agentthat optimizes the verified controller for performance in the same bounded setaround the control input. Third, we verify probabilistic safety guarantees withrespect to temporal logic specifications for the learned agent. Our approach isefficiently implementable for continuous action and state spaces. Theseparation of safety verification and performance improvement into two distinctsteps realizes both explicit probabilistic safety guarantees and astraightforward RL setup that focuses on performance. We evaluate our approachon an evasion task where a robot has to reach a goal while evading a dynamicobstacle with a specific maneuver. Our results show that our safe RL approachleads to efficient learning while maintaining its probabilistic safetyspecification.

Safe Reinforcement Learning with Probabilistic Guarantees Satisfying Temporal Logic Specifications in Continuous Action Spaces