In present-day Japan, the number of people with physical disabilities tends to rise year by year, with the proportion of those aged 65 and over also rising. In addition, Japan today is a super-aged society, and the number of people returning their driving licences is increasing year by year due to the decline in physical functions that accompanies ageing. These two factors led us to focus on research into electric wheelchairs suitable for the physically challenged and elderly to safely move around outdoors. The problems with conventional electric wheelchairs include the difficulty of learning driving techniques and accidents caused by malfunctions. The aim of this research was not only to solve the above problems by realising autonomous driving of electric wheelchairs, but also to construct a navigation system that enables passengers to move safely to any location. The electric wheelchair uses GPS and magnetic sensors to estimate the self-position and attitude of the aircraft and to perform navigation. In addition, by using an arbitrary point on a map prepared in advance as a checkpoint, and using CNN-based machine learning to landmark objects in the vicinity that are easily captured by the features, self-position estimation and navigation correction are carried out while realising autonomous driving of the aircraft.