The article introduces a novel dataset and experimental evaluation focusing on safe road-crossing by autonomous wheelchairs. It addresses the challenges of trustworthiness in AI-based systems, particularly in safety-critical scenarios. The work is part of the REXASI-PRO project aiming to develop reliable AI for social navigation. The study involves a system with an autonomous wheelchair and a drone equipped with diverse sensors. By combining artificial vision and distance sensors, the authors designed an analytical danger function to support real-time risk-based decision-making for road-crossing scenarios without traffic lights. Experimental evaluations in a laboratory environment demonstrated the benefits of using multiple sensors to enhance decision accuracy and safety assessment.
To Another Language
from source content
arxiv.org
Deeper Inquiries