In Senegal, blind and visually impaired people face many challenges, particularly in terms of accessibility and mobility. Thus, this research work proposes a smart cane based on Internet of Things (IoT) and artificial intelligence (AI) to assist visually impaired people in their movements. The proposed cane integrates several sensors, such as ultrasound, buzzer and a pi camera. An AI model was trained using an open-source software, YOLO, to analyze the nature of obstacles, such as cars, stones, person, puddles, chair, etc. This allows the model to identify and classify the obstacles encountered by the cane's holder. A GPS module is also integrated for the real-time location of the cane. The development of a web platform, the prototype deployment of the smart cane and case studies for validation show good AI accuracy in detection and precise location, which makes it easier for users of the cane to move around by informing them about their surroundings and guaranteeing them more safety.