Heading estimation algorithm for pedestrian navigation using a dual-foot-mounted IMU with a millimetre wave radar

•Single-foot multiconditional constraints, adaptive step size constraints, bipedal maximum heading difference constraints, and millimeter-wave radar-based inter-foot ranging constraints are fused into a bipedal multiconditional constraint algorithm.•Practical experiments were conducted to verify that the algorithm does not over-adjust the heading under smooth paths, as well as to verify that the algorithm under-corrects the heading at sharp turns.•The proposed heading estimation algorithm, which is suitable for very stealthy foot-tethered positioning systems, has great potential for commercial applications in the field of professional pedestrian positioning. Heading estimation is crucial in pedestrian navigation systems based on IMUs, but accuracy often suffers degradation due to error accumulation. This study introduces a novel heading estimation algorithm using dual-foot-mounted IMUs with a millimeter-wave radar to enhance accuracy. The algorithm includes a Single-foot Multiconditional Constraint Algorithm (SMC-A), adaptive step length constraint, and bipedal maximum heading difference constraint. Firstly, the SMC-A utilizes zero speed correction, zero angular rate updates, and self-observing heading error corrections through extended Kalman filtering. Then, the adaptive step length constraint limits pedestrian motion, and the bipedal constraint corrects heading errors. Finally, the Bipedal Multiconditional Constraint Algorithm (BMC-A) incorporates millimeter-wave radar measurements of foot distance. Experiments show heading deviations of 3.74° and 4.03° are obtained for the left and right feet over a 30-minute, 1.45 km trial. The new algorithm improves heading accuracy by 68.19 % and 70.67 % for the left and right feet, respectively, compared to traditional SMC-A.