Intelligent Secure Trustable Things

InSecTT Intelligent Secure Trustable Things, aimed at bringing the Internet of Things and Artificial Intelligence together to achieve the full potential of the “Artificial Intelligence of Things". InSecTT is a pan-European effort with 52 partners from 12 countries that provided intelligent, secure and trustworthy systems for industrial applications that are cost-efficient, end-to-end secure, trustworthy, interconnected, and interoperable, thereby brining the Internet of Things and Artificial Intelligence together.

Swedish InSecTT results include: (i) a platform for connected automated vehicles, (ii) end-to-end ML platform for preventive maintenance and anomaly detection, (iii) driver distraction monitoring/detection, (iv) two network traffic datasets, factory sensor, and lidar datasets, (v) ML techniques to detect network anomalies, (vi) enforcement architecture for dynamic access control, and (vii) federated learning for Network Attack Detection/Classification to ensure safe and secure manufacturing.

RISE coordinated the Swedish part of InSecTT, with additional partners ABB, KTH, MDU, RTE, TietoEvry, and Westermo, involved in the demonstrators: (i) platooning of connected vehicle with a minimal risk manoeuvre, (ii) (semi) autonomous vessel, (iii) driving distraction detection, (iv) anomaly detection in industry communication systems, and (v) modular Ice-cream factory. The Westermo operating system was extended to run applications in containers.

Use case on "Driver Monitoring and Distraction Detection using AI":  RISE contributed with a technical implementation and a demonstrator to address the topic of driver behaviour monitoring and distraction detection. These distractions are defined as distractions where drivers engage with devices or activities unrelated to driving for durations exceeding two seconds and these are linked, according to research, to doubling the risk of a collision. We monitor such device usage via: a) smartphone sensors, b) wearables (smartwatches) sensors and c) camera-vision for labelling the events. We use AI models on edge devices for the identification of distractions from the sensors and cameras. The recorded sensor data is then presented in a web application fusing information to educate and train drivers.

https://youtu.be/bZW1eEvL57k