Starting My Research Project Journey: Building a Sensor-Driven Digital Twin for Vehicle Safety

Welcome to this blog. This space will document my final-year research project in Computer Systems Engineering, where I explore how digital twin technology can be used to improve safety in autonomous and semi-autonomous vehicles.

Autonomous driving systems operate in safety-critical environments, where mistakes are costly and real-world testing is both risky and limited. One of the biggest challenges in this field is validating perception and decision-making systems reliably, especially when multiple sensors are involved. This project is driven by a simple but important question:

How can we safely and repeatably evaluate collision-prevention systems before deploying them in the real world?

To answer this, my research focuses on developing a sensor-driven digital twin a virtual representation of a vehicle’s perception system that is continuously updated using sensor data. The system combines camera-based computer vision, LiDAR distance sensing, and mmWave radar velocity measurement to detect obstacles and assess collision risk. A YOLO-based object detection model processes visual data, while LiDAR and radar provide accurate spatial and motion information. These inputs are fused to estimate Time-To-Collision (TTC), enabling early collision warnings.

What makes this project different is the role of the digital twin. Rather than treating simulation as a visual aid, the digital twin is used as a core validation and safety analysis tool. By mirroring sensor behaviour in real time, it allows collision-prevention logic to be tested under controlled, repeatable conditions—without the risks associated with real-world experiments.

Over the coming weeks, this blog will cover:

• System design decisions and architecture

• Sensor fusion challenges and solutions

• Digital twin validation experiments

• Performance evaluation and safety analysis

• Lessons learned from both simulation and hardware integration

This blog is intended as a technical research log, but also as a resource for anyone interested in autonomous systems, sensor fusion, and digital twin–based validation. If you’re curious about how virtual and physical systems can work together to improve safety, you’re in the right place.