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

Imitation learning has been widely used to learn control policies forautonomous driving based on pre-recorded data. However, imitation learningbased policies have been shown to be susceptible to compounding errors whenencountering states outside of the training distribution. Further, these agentshave been demonstrated to be easily exploitable by adversarial road usersaiming to create collisions. To overcome these shortcomings, we introduceAdversarial Mixture Density Networks (AMDN), which learns two distributionsfrom separate datasets. The first is a distribution of safe actions learnedfrom a dataset of naturalistic human driving. The second is a distributionrepresenting unsafe actions likely to lead to collision, learned from a datasetof collisions. During training, we leverage these two distributions to providean additional loss based on the similarity of the two distributions. Bypenalising the safe action distribution based on its similarity to the unsafeaction distribution when training on the collision dataset, a more robust andsafe control policy is obtained. We demonstrate the proposed AMDN approach in avehicle following use-case, and evaluate under naturalistic and adversarialtesting environments. We show that despite its simplicity, AMDN providessignificant benefits for the safety of the learned control policy, whencompared to pure imitation learning or standard mixture density networkapproaches.

Adversarial Mixture Density Networks: Learning to Drive Safely from Collision Data