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

We propose improving the cross-target and cross-scene generalization ofvisual navigation through learning an agent that is guided by conceiving thenext observations it expects to see. This is achieved by learning a variationalBayesian model, called NeoNav, which generates the next expected observations(NEO) conditioned on the current observations of the agent and the target view.Our generative model is learned through optimizing a variational objectiveencompassing two key designs. First, the latent distribution is conditioned oncurrent observations and the target view, leading to a model-based,target-driven navigation. Second, the latent space is modeled with a Mixture ofGaussians conditioned on the current observation and the next best action. Ouruse of mixture-of-posteriors prior effectively alleviates the issue ofover-regularized latent space, thus significantly boosting the modelgeneralization for new targets and in novel scenes. Moreover, the NEOgeneration models the forward dynamics of agent-environment interaction, whichimproves the quality of approximate inference and hence benefits dataefficiency. We have conducted extensive evaluations on both real-world andsynthetic benchmarks, and show that our model consistently outperforms thestate-of-the-art models in terms of success rate, data efficiency, andgeneralization.

NeoNav: Improving the Generalization of Visual Navigation via Generating Next Expected Observations