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

Behavior prediction in dynamic, multi-agent systems is an important problemin the context of self-driving cars, due to the complex representations andinteractions of road components, including moving agents (e.g. pedestrians andvehicles) and road context information (e.g. lanes, traffic lights). This paperintroduces VectorNet, a hierarchical graph neural network that first exploitsthe spatial locality of individual road components represented by vectors andthen models the high-order interactions among all components. In contrast tomost recent approaches, which render trajectories of moving agents and roadcontext information as bird-eye images and encode them with convolutionalneural networks (ConvNets), our approach operates on a vector representation.By operating on the vectorized high definition (HD) maps and agenttrajectories, we avoid lossy rendering and computationally intensive ConvNetencoding steps. To further boost VectorNet's capability in learning contextfeatures, we propose a novel auxiliary task to recover the randomly masked outmap entities and agent trajectories based on their context. We evaluateVectorNet on our in-house behavior prediction benchmark and the recentlyreleased Argoverse forecasting dataset. Our method achieves on par or betterperformance than the competitive rendering approach on both benchmarks whilesaving over 70% of the model parameters with an order of magnitude reduction inFLOPs. It also outperforms the state of the art on the Argoverse dataset.

VectorNet: Encoding HD Maps and Agent Dynamics from Vectorized Representation