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

Predicting the future motion of observed vehicles is a crucial enabler forsafe autonomous driving. The field of motion prediction has seen large progressrecently with State-of-the-Art (SotA) models achieving impressive results onlarge-scale public benchmarks. However, recent work revealed thatlearning-based methods are prone to predict off-road trajectories inchallenging scenarios. These can be created by perturbing existing scenarioswith additional turns in front of the target vehicle while the motion historyis left unchanged. We argue that this indicates that SotA models do notconsider the map information sufficiently and demonstrate how this can besolved, by representing model inputs and outputs in a Frenet frame defined bylane centreline sequences. To this end, we present a general wrapper thatleverages a Frenet representation of the scene and that can be applied to SotAmodels without changing their architecture. We demonstrate the effectiveness ofthis approach in a comprehensive benchmark using two SotA motion predictionmodels. Our experiments show that this reduces the off-road rate on challengingscenarios by more than 90\%, without sacrificing average performance.

Stay on Track: A Frenet Wrapper to Overcome Off-road Trajectories in Vehicle Motion Prediction