An Inertial Sequence Learning Framework for Vehicle Speed Estimation via Smartphone IMU

--Accurately estimating vehicle velocity via smart-phone is critical for mobile navigation and transportation. This paper introduces a cutting-edge framework for velocity estimation that incorporates temporal learning models, utilizing Inertial Measurement Unit (IMU) data and is supervised by Global Navigation Satellite System (GNSS) information. The framework employs a noise compensation network to fit the noise distribution between sensor measurements and actual motion, and a pose estimation network to align the coordinate systems of the phone and the vehicle. T o enhance the model's generalizability, a data augmentation technique that mimics various phone placements within the car is proposed. Moreover, a new loss function is designed to mitigate timestamp mismatches between GNSS and IMU signals, effectively aligning the signals and improving the velocity estimation accuracy. Finally, we implement a highly efficient prototype and conduct extensive experiments on a real-world crowdsourcing dataset, resulting in superior accuracy and efficiency. HE emergence of smartphone-based vehicular applications has revolutionized how drivers access and take advantage of mobile services. These applications offer a wide range of valuable features that enhance driving safety and convenience, such as real-time vehicle positioning, analysis of driving behavior, intelligent navigation assistance, and traffic status updates. According to statistics, in 2021, nearly 70% of drivers use mobile navigation apps like Gaode and Baidu Maps while driving (Figure 1 (a)) in China. Ride-hailing drivers, in particular, rely heavily on the positioning services provided by these mobile navigation apps to ensure accurate passenger pick-up and drop-off. Consequently, navigation app service providers, such as DiDi, Uber, and Amap, are dedicated to enhancing the precision of smartphone-based vehicle positioning, thereby improving the user experience. Typically, Global Navigation Satellite System (GNSS) information provides position [1]. However, the limitations of mobile phone hardware and complex urban environments can lead to signal degradation and even congestion, which challenges GNSS to provide a consistently stable signal over long periods of time, especially when the vehicle passes through densely built areas, tunnels, or underground parking facilities (Figure 1 (b)). The absence of satellite perception significantly hampers the driving experience, for instance, in subterranean parking lots where the provided location diverges considerably from the actual position, driver may encounter confusion and disorientation.