In order to achieve the ultimate goal of a robust application that can reliably generate pace notes for rally drivers, development of the application is broken down into manageable chunks equating to 4 prototypes. Each prototype builds upon the last to enhance functionality and improve the output. Find out more below.
Our
Journey
Initial concept planned and architecture explored
The 1st prototype is built. This proof-of-concept relies entirely off computer vision for recognising corners and the severity of the turn
Prototype 2 builds upon the 1st prototype. Version 2 is designed to incorporate data from the car (e.g accelerometer data and GPS) to build on the core functionality by improving corner recognition accuracy especially on non-asphalt surfaces. It's also designed to detect the distance between corners.
This prototype includes the introduction of nural networks to give the application contextual understanding and adaptive learning capabilities. Prototype 3 is planned to be able to detect obsticals (e.g poles, rocks etc.) that need to be flagged in the pace notes.
The 4th iteration leverages machine learning for dynamic adjustment. It is planned that this version will be able to generate pace notes in adverse weather conditions and adjust the notes for the current road conditions.