Trajectory Management and Analysis

new ideas 2021
- come back to motion prediction: https://arxiv.org/abs/1908.11472
- loop back on work by Laurent Boucaud and Mohsen Rezaie
Data structure for trajectories?
- dimention: 2D, 3D?
- time dimension: constant sampling rate?
Application data:
- GPS: potentially different sampling rate
- fixed point detection (eg, BT)
- mobile id detection (BT from mobile sensors such as smartphones (Y. Malinovskiy))
- tracking from video data: potential missing points (occlusion)
- uncertain positions (spatial and time)
Exceptions ??
Python bindings: inspiration from OpenCV?
Generic interface to different database engines, possibly with spatial extensions (eg PostgreSQL with PostGIS, SQLite with SpatiaLite)
Representation of clusters, motion patterns: prototypes, paths, Gaussian processes
- motion prediction from historical data / large datasets
Use of Catch for tests (https://github.com/philsquared/Catch): header only

Trajectory Filtering