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

Adversarial attacks on learning-based trajectory predictors have already beendemonstrated. However, there are still open questions about the effects ofperturbations on trajectory predictor inputs other than state histories, andhow these attacks impact downstream planning and control. In this paper, weconduct a sensitivity analysis on two trajectory prediction models,Trajectron++ and AgentFormer. We observe that between all inputs, almost all ofthe perturbation sensitivities for Trajectron++ lie only within the most recentstate history time point, while perturbation sensitivities for AgentFormer arespread across state histories over time. We additionally demonstrate that,despite dominant sensitivity on state history perturbations, an undetectableimage map perturbation made with the Fast Gradient Sign Method can induce largeprediction error increases in both models. Even though image maps maycontribute slightly to the prediction output of both models, this resultreveals that rather than being robust to adversarial image perturbations,trajectory predictors are susceptible to image attacks. Using anoptimization-based planner and example perturbations crafted from sensitivityresults, we show how this vulnerability can cause a vehicle to come to a suddenstop from moderate driving speeds.

Hacking Predictors Means Hacking Cars: Using Sensitivity Analysis to Identify Trajectory Prediction Vulnerabilities for Autonomous Driving Security