<p>The availability of computationally efficient hardware has made real-time object detection very popular for numerous applications in various domains. Real-time staircase detection is one such problem, with tremendous applications in robotics, self-moving machines and indoor vehicles. The evolution of new Artificial Neural Network (ANN) and machine learning (ML) techniques has unlocked innovative ways to implement more advanced algorithms for real-time staircase detection. You Only Look Once (YOLO) is a popular real-time object detection algorithm that makes the predictions of bounding boxes and class probabilities all at once using a single neural network. YOLOv5 is the popular release of the You Only Look Once (YOLO) series, which can be implemented on the Jetson nano board for the detection of the staircase in real-time. This paper discusses the implementation of real-time staircase detection on the Jetson nano board using the YOLOv5 model. The YOLOv5 model is trained on the staircase dataset for the detection of four different types of staircases namely ascending staircase, descending staircase, ascending curvature staircase and descending curvature staircase. The proposed model obtained the mean average precision accuracy of 0.953 during the training on the staircase dataset. We have obtained higher precision and recall values for real-time staircase detection. We have successfully detected the staircase in real-time with 60 fps using Jetson Nano.</p>