Enabling Full-State Observability in Pedestrian Localization via Active Kinematic Excitation

Pedestrian Inertial Navigation Systems (PINS) enable ubiquitous localization for the Internet of Things (IoT), particularly in GPS-denied environments. However, low-cost inertial sensors suffer from significant error accumulation. While the Zero-Velocity Update (ZUPT) is widely adopted to constrain this drift, the standard ZUPT framework struggles to estimate sensor biases effectively. This limitation stems from the high repeatability of foot attitude during the stance phase, which provides insufficient kinematic excitation. Consequently, critical bias states remains weakly observable, leading to rapid positioning drift. To break the observability bottleneck, this paper proposes a novel gait-coupled discrete rotation method. Unlike complex continuous rotation systems, we utilize the swing phase to actively change the IMU attitude. A quantitative observability analysis framework is established to derive a mathematically optimal rotation sequence. To validate this approach, a prototype is constructed using consumer-grade IMUs. The comprehensive 30-minute 2-kilometer walking test shows an end-to-end precision of 1.35%D, demonstrating a 62.47% improvement over conventional configurations. This work confirms that active kinematic excitation significantly enhances low-cost PINS performance, offering a new paradigm for affordable high-precision localization.