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

Self-driving cars and trucks, autonomous vehicles (AVs), should not beaccepted by regulatory bodies and the public until they have much higherconfidence in their safety and reliability -- which can most practically andconvincingly be achieved by testing. But existing testing methods areinadequate for checking the end-to-end behaviors of AV controllers againstcomplex, real-world corner cases involving interactions with multipleindependent agents such as pedestrians and human-driven vehicles. Whiletest-driving AVs on streets and highways fails to capture many rare events,existing simulation-based testing methods mainly focus on simple scenarios anddo not scale well for complex driving situations that require sophisticatedawareness of the surroundings. To address these limitations, we propose a newfuzz testing technique, called AutoFuzz, which can leverage widely-used AVsimulators' API grammars to generate semantically and temporally valid complexdriving scenarios (sequences of scenes). To efficiently search for trafficviolations-inducing scenarios in a large search space, we propose a constrainedneural network (NN) evolutionary search method to optimize AutoFuzz. Evaluationof our prototype on one state-of-the-art learning-based controller, tworule-based controllers, and one industrial-grade controller in five scenariosshows that AutoFuzz efficiently finds hundreds of traffic violations inhigh-fidelity simulation environments. For each scenario, AutoFuzz can find onaverage 10-39% more unique traffic violations than the best-performing baselinemethod. Further, fine-tuning the learning-based controller with the trafficviolations found by AutoFuzz successfully reduced the traffic violations foundin the new version of the AV controller software.

Neural Network Guided Evolutionary Fuzzing for Finding Traffic Violations of Autonomous Vehicles