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

In this work, we present a learning-based pipeline to realise localnavigation with a quadrupedal robot in cluttered environments with static anddynamic obstacles. Given high-level navigation commands, the robot is able tosafely locomote to a target location based on frames from a depth camerawithout any explicit mapping of the environment. First, the sequence of imagesand the current trajectory of the camera are fused to form a model of the worldusing state representation learning. The output of this lightweight module isthen directly fed into a target-reaching and obstacle-avoiding policy trainedwith reinforcement learning. We show that decoupling the pipeline into thesecomponents results in a sample efficient policy learning stage that can befully trained in simulation in just a dozen minutes. The key part is the staterepresentation, which is trained to not only estimate the hidden state of theworld in an unsupervised fashion, but also helps bridging the reality gap,enabling successful sim-to-real transfer. In our experiments with thequadrupedal robot ANYmal in simulation and in reality, we show that our systemcan handle noisy depth images, avoid dynamic obstacles unseen during training,and is endowed with local spatial awareness.

Learning a State Representation and Navigation in Cluttered and Dynamic Environments