• autonomous driving
• simulation
• simulator
• assisted driving
• assistive technologies
• power wheelchair
• artificial intelligence
• machine learning
• ai

Many power wheelchair users have severe physical disabilities affecting also the upper limbs, depriving them to drive safely or react promptly in several critical situations.
In the last decade several research projects have tried to improve the safety in the power wheelchair by exploiting the recent advances in mobile robotics and autonomous vehicles.
Unfortunately, none of these research prototypes have reached the market mainly due to the high
cost involved.

On the contrary, the solution proposed in this work exploits low-cost sensors combined with innovative Machine Learning algorithms to provide the required information.

The thesis work is instrumental to the development of a smart wheelchair enhanced with a semi-autonomous assisted driving system, working in both indoor and outdoor scenarios without prior knowledge of the environment.
The power wheelchair is capable of detecting various harmful situations as (i) the collision with static and dynamic objects of different shape and size, (ii) the risk of rollover and (iii) the uncomfortable high accelerations;
Moreover, the proposed system leaves to the end-user the maximum freedom in driving the wheelchair and the final decision to activate or not the semi-autonomous assisted driving mode.