Fabian Jarmolowitz | Searcharxiv

arXiv2024

Vehicle single track modeling using physics guided neural differential equations

In this paper, we follow the physics guided modeling approach and integrate a neural differential equation network into the physical structure of a vehicle single track model. By relying on the kinematic relations of the single track ordinary differential equations (ODE), a small neural network and few training samples are sufficient to substantially improve the model accuracy compared with a pure physics based vehicle single track model. To be more precise, the sum of squared error is reduced by 68% in the considered scenario. In addition, it is demonstrated that the prediction capabilities of the physics guided neural ODE model are superior compared with a pure black box neural differential equation approach.

cs.CE↗

arXiv2021

Detection of Cyber-Attacks in Collaborative Intersection Control

Road intersections are widely recognized as a lead cause for accidents and traffic delays. In a future scenario with a significant adoption of Cooperative Autonomous Vehicles, solutions based on fully automatic, signage-less Intersection Control would become viable. Such a solution, however, requires communication between vehicles and, possibly, the infrastructure over wireless networks. This increases the attack surface available to a malicious actor, which could lead to dangerous situations. In this paper, we address the safety of Intersection Control algorithms, and design a Sliding-Mode-Observer based solution capable of detecting and estimating false data injection attacks affecting vehicles' communication. With respect to previous literature, a novel detection logic with improved detection performances is presented. Simulation results are provided to show the effectiveness of the proposed approach.

eess.SY↗