In indoor environments where satellite signals are obstructed, inertial navigation technology demonstrates a high level of autonomy and interference resistance. Compared to other pedestrian inertial positioning systems, foot-mounted inertial navigation systems exhibit distinct gait characteristics and periods of stasis, offering favorable conditions for error correction in low-cost micro-electro-mechanical system inertial positioning systems, which suggests a broad range of applications. However, current gait detection thresholds are static, the determination methods are singular, interval division is ambiguous, and there are significant issues with missed and false detections. Moreover, inertial navigation algorithms lack positional information observation, and without external information assistance, the positioning results tend to diverge over time. Existing step length models that constrain position have fixed parameters, leading to poor adaptability. To address these challenges, this paper introduces a foot-mounted inertial navigation algorithm that is based on dynamic threshold gait detection and adaptive step length estimation. Initially, a dynamic threshold gait detection method is proposed, which integrates multi-condition detection based on the transformation rules of foot motion information. Subsequently, leveraging the high short-term precision of inertial navigation algorithms, the linear step length estimation model is refined to establish an adaptive step length estimation model with self-regulating parameters. Ultimately, the estimated single-step displacement of the foot is utilized to estimate positional information as an observation, thereby correcting the positional errors of the foot-mounted system. In multiple 1000 m open-loop tests, the algorithm presented in this paper achieved a maximum endpoint error of 12.54 m, a mean error of 11.68 m, and a root mean square error of 11.70 m.

The pedestrian inertial navigation systems are generally based on Pedestrian Dead Reckoning (PDR) algorithm. Considering the physiological characteristics of pedestrian movement, we use the cyclical characteristics and statistics of acceleration waveform and features which are associated with the walking speed to estimate the stride length. Due to the randomness of the pedestrian hand-held habit, the step events cannot always be detected by using the periods of zero velocity updates (ZUPTs). Furthermore, the signal patterns of the sensor could differ significantly depending on the carrying modes and the user's hand motion. Hence, the step detection and the associated adaptive step length model using a handheld device equipped with accelerometer is required to obtain high-accurate measurements. To achieve this goal, a compositional algorithm of empirical formula and Back-propagation neural network by using handheld devices is proposed to estimate the step length, as well as achieve the accuracy of step detection higher than 98%. Furthermore, the proposed joint step detection and adaptive step length estimation algorithm can help much in the development of Pedestrian Navigation Devices (PNDs) based on the handheld inertial sensors. Index Terms—Pedestrian inertial navigation; handheld devices; step detection; step length estimation; accelerometer

—The demand for navigating a user with a hand-held device, especially in Global Position System (GPS) denied environments, has tremendously increased over the last few years. Accelerometers, gyroscopes, and magnetometers are the most commonly found sensors in the smartphones that provide Three Dimensional (3D) acceleration and attitude of the phone. Algorithm of pedestrian navigation with smart phones modes switching is studied. When the sensor is rigidly mounted on the user’s body, the trajectory of the user can easily be reconstructed. The placement of the phone can vary overtime as a user performs different tasks. When the sensor’s location is dynamically changing, the situation becomes much more complex. Smartphone modes among three most commonly used are considered in this research, texting mode, ear-talking mode and waist mode. Using the machine learning method of decision trees is developed to recognize smartphones’ modes. The average accuracy of the selected classifier is > 92.8%. According to the detected smartphone mode, adaptive heading angle compensation algorithms are applied, the location error in the horizontal direction from the starting point to the ending point is approximately less than 30 meters when people with smartphone mode switched walk a distance of 1000 meters, and the feasibility of the algorithm is verified. The dynamic measurement precision of pedestrian navigation using a smart phone is improved, and it is more accurate to use a smart phone to realize pedestrian navigation in different smartphone modes.