Mobile Sensor Path Planning for Kalman Filter Spatiotemporal Estimation

Area of Science:

• Scientific disciplines requiring spatiotemporal data estimation.
• Sensor networks and data assimilation.

Background:

• Estimating spatiotemporal data from limited sensor measurements is crucial across sciences.
• Kalman filtering is a key technique for data estimation, balancing model and measurement data.

Purpose of the Study:

• To optimize sensor placement and data estimation using mobile sensors and Kalman filtering.
• To develop a scalable and computationally efficient greedy path planning algorithm for mobile sensing.

Main Methods:

• Utilizing greedy algorithms and low-rank subspace projection for model-free sensor selection.
• Applying Kalman filtering to integrate historical and current measurements from mobile sensors.
• Developing a greedy path planning algorithm based on minimizing the condition number of the observability matrix.

Main Results:

• Mobile sensing along dynamic trajectories achieves performance equivalent to a larger number of stationary sensors.
• Performance gains are influenced by spatiotemporal dynamics, sensor velocity, and sampling rate.
• The proposed path planning algorithm demonstrates improved scalability and computational efficiency.

Conclusions:

• Mobile sensing, particularly along optimized dynamic trajectories, significantly enhances Kalman filter performance for spatiotemporal data estimation.
• The method is effective for capturing spatially localized structures in dynamic datasets.
• The approach offers a more efficient and scalable solution compared to previous methods.