Tensor-field based localization

Anomalies in the earth’s magnetic and gravity field constitute an omnipresent and robust source for localization in GPS denied environments. Today, accelerometer and magnetometer sensor arrays can be constructed and integrated into portable platforms.

Similarly as a camera can take an image of the surrounding, these arrays can take an image like measurement of a field referred to as a tensor-field observation.

And just as in vision based localization, these tensor-field observation can be used for odometry, absolute localization, and simultaneous localization and mapping (SLAM).The PI, the co-PI, and a PhD student, will over 4-year period use estimation theoretical and system identification tools to research techniques for self-localization using tensor-field observations.

Focus will be on developing sensor-fusion, vector-field modeling, and model-learning techniques that can enable significantly longer stand-alone operation time of inertial navigation systems, and boost the accuracy and scalability of the state-of-the-art techniques for vector-field based SLAM.

Access to “anytime and anywhere” localization is frequently taken for granted in research connected to intelligent and autonomous systems. However, current localization technologies often fail to meet these demands.

Therefore, the project will be of importance to bridge the gap between today´s localization technology and the localization capabilities needed to design tomorrow´s intelligent and autonomous systems.