Neural Inertial Navigation
CONTEXT
The Inertial Measurement Unit (IMU) consists of the triplet of an Accelerometer, a Gyroscope, and a Magnetometer to measure changes in terms of linear velocity, angular velocity, and magnetic field, respectively. This unit holds a core component in enhanced navigation and recreational applications, e.g., Autonomous Cars, Augmented Reality, and Virtual Reality. Inertial Navigation is meant to utilize the sensitivity of the IMU as a sensing unit to estimate the translation and rotation offsets from movements for position heading computation. Due to variabilities caused by human and environmental conditions, the traditional methods started with Strapdown Inertial Navigation Systems based on mathematical integration, and Pedestrian Dead-Reckoning (PDR) variants with means of step detection and stride length estimation blocks both for position estimation, show more their limited performance in operation with accrued drift, which remarkably deviates the estimated trajectory from the expected one. In response to these challenges, data-driven methods which are able to gain more insights into the dynamic environment have been proposed to reduce the drift and to efficiently and independently work without any sub-blocks.
Task
In this work, students are expected to fully review state-of-the-art work around this research and then carefully analyze the strengths and weaknesses of respective ones. From that basis, they are positively encouraged to propose their own novel solutions to error reduction, and reliability enhancement.
YOU WILL GET
- Profound experience in related fields, such as deep learning models and their applications, plus time-series data properties, particularly IMU to movements and environments.
- A publication at top-tier AI venues if the work is qualified.
REQUIREMENTS
- Strong background in Machine Learning and Deep Learning.
- At least familiar with one of the AI frameworks, such as Tensorflow, and Pytorch.
- Having hands-on experience in basic Deep Learning projects.
- A good comprehension of related papers.
Contact:
Minh Son Nguyen, m.s.nguyen@utwente.nl