Detection and rehabilitation treatments of neuro-degenerative diseases through virtual reality are increasingly growing in technology and development in modern times. The PEDAL project aims to provide support for treatments of patients with cognitive diseases such as Alzheimer's by combining user immersion in a realistic virtual environment with movements on a stationary bicycle controlling their own virtual avatar. This paper describes methods for defining orientation paths within realistic city neighbourhoods, aimed at being proposed as an exercise for memory and spatial orientation. Discerning the well-known path finding problems that we find in robotics and gaming, an algorithm is developed to detect a realistic-looking path leading the user from a certain location to their virtual home.