| Description |
The need for position and orientation information in a wide variety of applications has led to the development of equally varied methods for providing it. Amongst the alternatives, inertial navigation is a solution that o ffers self-contained operation and provides angular rate, orientation, acceleration, velocity, and position information. Until recently, the size, cost, and weight of inertial sensors has limited their use to vehicles with relatively large payload capacities and instrumentation budgets. However, the development of microelectromechanical system (MEMS) inertial sensors now o ers the possibility of using inertial measurement in smaller, even human-scale, applications. Though much progress has been made toward this goal, there are still many obstacles. While operating independently from any outside reference, inertial measurement su ers from unbounded errors that grow at rates up to cubic in time. Since the reduced size and cost of these new miniaturized sensors comes at the expense of accuracy and stability, the problem of error accumulation becomes more acute. Nevertheless, researchers have demonstrated that useful results can be obtained in real-world applications. The research presented herein provides several contributions to the development of human-scale inertial navigation. A calibration technique allowing complex sensor models to be identified using inexpensive hardware and linear solution techniques has been developed. This is shown to provide significant improvements in the accuracy of the calibrated outputs from MEMS inertial sensors. Error correction algorithms based on easily identifiable characteristics of the sensor outputs have also been developed. These are demonstrated in both one- and three-dimensional navigation. The results show significant improvements in the levels of accuracy that can be obtained using these inexpensive sensors. The algorithms also eliminate empirical, application-specific simplifications and heuristics, upon which many existing techniques have depended, and make inertial navigation a more viable solution for tracking the motion around us. |
