Spatially Separated Redundant Magnetic Sensors and Nonlinear Observers for Enhanced Position Estimation

This award supports the investigation of a novel sensing principle for estimation of positions of ferromagnetic objects through a combination of nature-inspired ideas. The spatial separation of identical sensors for enhanced sensory perception (e.g. dual eyes and ears in animals and humans) will be exploited along with the inherent magnetic fields of objects for determining their position as moving objects. The new sensing principle will enable non-contact measurement of position without requiring the placement of any sensors or sensing components on the ferrous object. Moving ferromagnetic objects such as cars, pistons, spool valves and a large variety of moving components in various machines have magnetic fields which vary as a function of the position around the object. The spatial variation of an object's magnetic field as a function of position, the measurement of magnetic field at a few discrete spatially separated locations and the use of novel estimation algorithms yield accurate estimation of the position of the monitored objects, with no pre-calibration required. This technology would have extensive applications in automotive, industrial and off-road vehicle domains.

In order to exploit the 'inherent' magnetic fields of common ferromagnetic objects, a number of technical challenges need to be addressed. These include unknown magnetic field parameters, the need to design nonlinear observers and the need for estimation techniques that can account for constraints on states and parameters. These challenges are addressed by using dual sensors with a known separation distance between them and by formulating rigorous observer design methods for nonlinear systems. The use of redundant sensors enables observability and enables estimation of both parameters and states without requiring pre-calibration. The design of nonlinear observers provides global guarantees of stability, is computationally efficient, elegant, and yields faster convergence. Constraints on states or parameters, and transient errors due to unexpected magnetic disturbances, are addressed using a moving horizon estimation method in which known bounds are imposed using a constrained optimization formulation. Major project tasks include analytical study of nonlinear observer design methods, development of disturbance rejection techniques, and implementation of constraint imposition techniques. These will be applied for position estimation for a free-piston engine, a hydraulic cylinder, and an automotive imminent collision detection.