Technology Impact Cycle Tool

Rekeningrijden

Computer vision

AI-Assisted Visual Inspection PoC for PCUs This technology is a proof-of-concept computer vision system designed to support visual inspection of PCUs on the manufacturing shop floor. A fixed camera captures an image of each PCU under defined conditions. The image is analysed by an anomaly-detection model that has learned the visual appearance of normal reference PCUs. The system produces an anomaly score and a localised heatmap to indicate image regions that differ from the normal-reference data. The intended purpose is to help operators identify visually observable deviations earlier and direct their attention to areas that may require closer inspection. The system may support more consistent visual inspection and provide additional visual evidence for quality review. It is not designed to identify the root cause of a deviation, classify a specific defect type, or autonomously decide whether a PCU should be accepted, reworked, or rejected. The final quality decision remains the responsibility of the operator and authorised quality personnel. The technology addresses the limitation of manual visual inspection, where small or subtle deviations may be difficult to notice consistently and inspection outcomes can depend on individual experience, workload, and environmental conditions. By comparing a captured image with normal-reference images, the system can highlight visual differences that could otherwise receive less attention. However, the expected benefits remain conditional. The PoC was evaluated only under defined test conditions, including a fixed camera configuration, controlled lighting, a limited set of normal PCUs, and selected introduced visual deviations. The results show that the system can generate anomaly scores and heatmaps for these cases. They do not yet demonstrate reliable performance across all PCU variants, real production defects, environmental variation, or long-term shop-floor use. A larger validation study with representative production data and continued human oversight would be required before operational deployment.

Believable NPCs game pacakge

A game package that students can put into their to add believable NPCs. The system will be very customizable and extendible so that it can be applied to a broad spectrum of student games.

Freelance Tax Manager

Track your income, manage clients, and stay on top of your tax obligations - all in one place.

Website

Gym motivational website that helps users keep each other accountable. It functions by each user adding a goal where they set a target for times per week to go to the gym and they join a group together. Each group has a bet/stake (common for everyone in the same group) and if a person does not fulfill the target they set themselves, they must pay the bet/stake to the other members of the group that fulfilled their goals!