The objective of sensor fusion is to optimally interconnect information from different sensors, and in the process take knowledge of occurring measuring errors into account. Frequently Bayes estimation methods are used in the investigation of dynamic processes. In this regard, any preliminary knowledge concerning the dynamics of a process and ambient parameters is included in the estimate. Moreover, it is possible to estimate variables that cannot be directly observed via a sensor. If real-time processing is necessary, then recursive Bayes estimation methods can be used, with which the variable to be estimated is represented through a probability distribution that is updated with each time step.
One example for an application is locating a person within a building using radio-based distance measurements, inertial sensors, and map information. Although the distance data is highly disturbed through reflections on walls, and inertial sensors alone are not suitable for locating due to drift, the combination of both concepts enables an amazing level of precision. With these methods, the orientation of individual body parts can also be determined or inferences are possible concerning the activities that the person is executing at the moment. Another example is improving the accuracy of an inclinometer through connection of a gyroscope. In this manner measuring errors that occur due to acceleration, such as occur when starting up and braking an automobile, can be compensated. Our services extend from feasibility studies to system concept and algorithm development, and extend to development and evaluation of prototypes.