DrTiago Tamagusko

My research asks how cities can count and assess walking and cycling as routinely as they already count cars. Motor traffic has been counted automatically for decades, while pedestrians and cyclists are still often counted by hand. I work on this at two levels. For measurement, I am developing edge-AI sensors that detect and count pedestrians, cyclists, e-scooters and vehicles on the device, without transmitting images (CAMINA), and I combine camera, LiDAR and GIS data to estimate exposure. For risk, I use vision-language models to assess cycling routes and children's school commutes from street-level imagery (COLOURWAYS). My doctoral work on machine learning for pavement performance continues alongside this, using transfer learning and synthetic data where agency records are limited. The aim throughout is methods that transport agencies can adopt and run at low cost.