E

edge computing

PulseDrive is an open-source, AI-assisted driver safety system that verifies sobriety before vehicle startup and continuously monitors driver identity and attention during operation. It combines fuel-cell breath analysis, CO₂ and pressure-based breath validation, and edge AI vision systems using dual RGB/IR cameras for driver monitoring and liveness detection. The system runs on edge hardware (such as NVIDIA Jetson or Intel NPU devices) and integrates with vehicles through a safe, read-only CAN/OBD-II interface that only authorizes ignition when safety conditions are met. PulseDrive is designed with a strict separation between deterministic safety-critical logic and advisory AI modules, emphasizing privacy, modularity, and automotive-grade safety principles under an AGPL-3.0+ open-source license. https://roxanneardary.com/pulsedrive/

ValidationOS is an open-source, local-first AI assurance platform that transforms AI performance into proven results. It provides tools for validating, optimizing, certifying, and monitoring AI models across edge hardware environments through automated testing, compatibility analysis, performance benchmarking, observability, and compliance workflows.

RoadCraft is an open-source AI motorcycle copilot system that combines helmet-based intelligence, real-time sensor fusion, and vehicle control integration to enhance rider safety without removing human control. It provides optional AI assistance for navigation, hazard detection, and emergency intervention, while supporting advanced features like V2X communication, stabilization systems, and edge-based processing. Built under the AGPL-3.0+ license, RoadCraft is designed for collaborative development of next-generation intelligent riding systems.

RequestStamp is a self-hosted edge request identity system that assigns a unique, traceable ID to every request entering your infrastructure. It enables real-time request tracking, rule-based decisioning, and full explainability of edge behavior across distributed systems. Designed for transparency and control, RequestStamp helps developers understand exactly what happened to any request, where, and why.

A real-time browser-based edge computing system that detects motion and intelligently captures only meaningful video events using on-device processing.

MSc degree thesis: multi-agent reinforcement learning for workload distribution in FaaS edge computing systems

Eclipse IoT Days is an annual event on IoT, Edge Computing and AI at Edge in Grenoble, France.

Expose your MicroK8s homelab to the Internet: https://www.donaldsebleung.com/blog/20240213-running-workloads-at-the-edge-with-microk8s

This project develops video analytics towards anomaly detection on edge for smart cities under low light conditions: https://iotgarage.net/projects/VideoAnalyticstowardsAnomalyDetectionontheEdge.html

Implementation of a Keyword Spotting Algorithm (KWS) using deep neural networks in the edge.

This project aims to develop a 100 node testbed to support data science and edge computing research. https://iotgarage.net/projects/LivingEdgeLaboratory.html