Abstract/Summary

The increasing number of vehicles on the road goes hand in hand with a rising number of traffic related accidents. As such there is a requirement to increase the level of safety provided to both the driver and other road users alike. The ultimate goal would be to realise autonomous vehicles on today’s roads. However few companies are able to provide a fully-fledged autonomous vehicle. That said many manufacturers are able to supply various advanced driver assistance systems in order to increase both the safety level and performance of the vehicle. Currently many different assistance systems are available that utilise a range of sensors, with great variation in cost. Not all vehicles have all sensors fitted, though there has been wide market acceptance of the ultrasonic parking aid. This sensor is low cost and also benefits from an ease of retrofitting, where not fitted as standard. Ultrasonic sensors however can only provide short range distance measurement and as such their uses are limited. This thesis takes ideas from CTFM ultrasound used in mobile robotics and applies them to low costs ultrasonic sensors found in the automotive industry, with the intention to provide additional functionality. In particular the specific challenge of driving surface recognition is considered. The experimental methods for data collection are presented alongside the classification techniques used. This thesis shows that low cost narrowband ultrasonic transducers can be used as an input for the classification of driving surfaces.