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 Plus annual

Presents the access of premium content as premium feature

Premium Content

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 Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Visual navigation is essential for many applications in robotics, frommanipulation, through mobile robotics to automated driving. Deep reinforcementlearning (DRL) provides an elegant map-free approach integrating imageprocessing, localization, and planning in one module, which can be trained andtherefore optimized for a given environment. However, to date, DRL-based visualnavigation was validated exclusively in simulation, where the simulatorprovides information that is not available in the real world, e.g., the robot'sposition or image segmentation masks. This precludes the use of the learnedpolicy on a real robot. Therefore, we propose a novel approach that enables adirect deployment of the trained policy on real robots. We have designed visualauxiliary tasks, a tailored reward scheme, and a new powerful simulator tofacilitate domain randomization. The policy is fine-tuned on images collectedfrom real-world environments. We have evaluated the method on a mobile robot ina real office environment. The training took ~30 hours on a single GPU. In 30navigation experiments, the robot reached a 0.3-meter neighborhood of the goalin more than 86.7% of cases. This result makes the proposed method directlyapplicable to tasks like mobile manipulation.

Visual Navigation in Real-World Indoor Environments Using End-to-End Deep Reinforcement Learning