Project ZaVI -State Estimation Solely Based on Inertial Sensors and Prior Knowledge
Inertial Navigation Systems (INS) are commonly used to track the pose (position and orientation) of an object. But the pose estimate of an INS alone does drift away from the true pose with time. Thus, they are often complemented by a Global Navigation Satelite Systems (GNSS) as the GPS. In certain environments, GNSS are unavailable. This would result in an increasing error of the pose estimate which yields the pose tracker unusable.
In the Project ZaVI we try to correct the drift of inertial sensors, by fusing the sensors with prior knowledge about the object. In general, the tracked object is subject to several motion and environmental constraints. These constraints allow to observe states of the object that would drift otherwise. The goal of the project is to understand how the constraints affect the observability of states.
The Project is funded by the German Research Federation.
Ongoing Research
We are currently investigating the prior knowledge available at bouldering. https://up2date.uni-bremen.de/en/research/climbing-with-sensors
Datasets
Our recorded IMU datasets from Track Cycling and Bouldering are available at: http://www.informatik.uni-bremen.de/zavi-datasets/info.html
Code Repositories: https://github.com/TomLKoller/
Publications
Tom Koller, Tim Laue, Udo Frese (2019). State Observability through Prior Knowledge: A Conceptional Paradigm in Inertial Sensing. In Oleg Gusikhin (Ed.), Proceedings of the 16th International Conference on Informatics in Control, Automation and Robotics, 1 (781-788). SCITEPRESS. Presented at ICINCO 19 http://www.icinco.org/ Icinco19.pdf
Tom Koller, Udo Frese (2019). State Observability through Prior Knowledge: Tracking Track Cyclers with Inertial Sensors. In International Conference on Indoor Positioning and Indoor Navigation IPIN at Pisa, Italy. IPIN.pdf
Tom Koller, Udo Frese (2020). State Observability through Prior Knowledge: Analysis of the Height Map Prior for Track Cycling. In MDPI Sensors Special Issue on Sensors and Sensing Technologies for Indoor Positioning and Indoor Navigation. MDPI
