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

In recent years, Deep Reinforcement Learning (DRL) has emerged as a promisingmethod for robot collision avoidance. However, such DRL models often come withlimitations, such as adapting effectively to structured environments containingvarious pedestrians. In order to solve this difficulty, previous research hasattempted a few approaches, including training an end-to-end solution byintegrating a waypoint planner with DRL and developing a multimodal solution tomitigate the drawbacks of the DRL model. However, these approaches haveencountered several issues, including slow training times, scalabilitychallenges, and poor coordination among different models. To address thesechallenges, this paper introduces a novel approach called evolutionarycurriculum training to tackle these challenges. The primary goal ofevolutionary curriculum training is to evaluate the collision avoidance model'scompetency in various scenarios and create curricula to enhance itsinsufficient skills. The paper introduces an innovative evaluation technique toassess the DRL model's performance in navigating structured maps and avoidingdynamic obstacles. Additionally, an evolutionary training environment generatesall the curriculum to improve the DRL model's inadequate skills tested in theprevious evaluation. We benchmark the performance of our model across fivestructured environments to validate the hypothesis that this evolutionarytraining environment leads to a higher success rate and a lower average numberof collisions. Further details and results at our project website.

Evolutionary Curriculum Training for DRL-Based Navigation Systems