Page 37 - Demo

35Filling the gaps: AI-powered dead reckoning (DR) modelDead reckoning (DR) is a way to calculate the current position of a moving object by adding the distance it has travelled to its previously determined position. On smartphones, DR is measured by IMUs and is often used to fill the gaps between GNSS positions in urban settings. However, IMU data on smartphones are of poor quality and can result in major errors. To make IMU data on smartphones more useful, Prof. Chen and his team developed a new DR model based on deep learning. This model is trained with a vast volume of positioning data from smartphones and significantly improves the accuracy of DR calculations in smartphones. When combined with Bluetooth and wi-fi, DR can also be integrated for efficient indoor navigation.Vertical navigation: smartphone barometersSome smart city applications, such as drones that monitor the structural health of buildings and surveillance robots that move inside buildings, require height information. To determine on which level a device is located within a building, the team used Hong Kong atmospheric pressure data to calibrate the inaccuracies that are inherent in smartphone barometers. They found that they could determine absolute height to an accuracy of within one metre, making this method useful for indoor navigation.Prof. Chen%u2019s research seamlessly connects precise indoor and outdoor positioning. This not only makes life easier but also paves the way for smart city applications and a brighter future. Each small step that Prof. Chen has made in achieving positioning precision marks a giant leap for mankind.ON THE BUTTON