Actions¶

Concerning DFNs¶

A.: Fault detection in PoM: this comes down to data interpretation. Each time a pose is computed, an associated covariance is produced. The question is: how can the covariance (or any measure of error) be used to detect when a localization module is being faulty?
Leads: outlier covariance? Slippage detection?

A.: Fault detection in DEM building: Each time a new Point cloud is available, a local DEM is built. In order for it to be fused with a larger DEM, we have to make sure that the localization of the robot is precise enough, else we are going to fuse two different geographic area and add noise to the DEM.
Leads: outlier height could be detected? Quadratic error measure between local DEM and internal/global DEM?

Data types definition and taxonomy: Lots of datatypes are already available in the rock core datatypes, ASN.1 format. However some of our needed datatypes do not exist yet and have to be defined.
Leads: create a list of all datatypes yet to be defined and start producing corresponding ASN.1 files.

DPM data storing:
- DPM: raw data, intermediary products, data products- cf ORB Slam
- Prepare a development plan and a validation plan for LAAS robots > in boxes
sujets de stage ? > TBD (different DFIs of the same DFNs, e.g. test different visual extractors)
Choisir la liste des capteurs, se faire 2 super robots (caméra panoramique ?) ->