Abstract layers for dealing with perceptive models framework. More...
Collaboration diagram for perceptiveModels:
Modules | |
| Interfaces | |
| A collection of abstract classes establishing the perceptive-based framework. | |
| Implementations | |
| Herein are grouped up all the perceptive models that rely on the framework. | |
Detailed Description
Abstract layers for dealing with perceptive models framework.
Description
This library provides the user with a collection of objects in order to organically handle the problem of acquiring data from the robot, execute some first grade processing and then pass them to the higher software layers.
The final objective is to give the raw data a kind of common structure together with the set of operations that are normally carried out at a preliminary stage such as the normalization of the values, the data formatting, the gathering of the data from different sources so that it turns to be convenient to model this structure to then apply alternative algorithms/techniques decoupling the higher routines from the acquisition layer.
As practical example, one may think to address the problem of sensing external contacts with the fingers of the robot either by relying on a elastic model of the distal links or by directly reading the output of the tactile sensors. This is only a mere implementative aspect (yet quite significant indeed) from the standpoint of a software designer, thus it can be relevant to build a sort of wrapper over the acquisition phase that in turn returns an homogeneous measure of the amount of contact regardless the source of the raw data. This is exactly what this library attempts to do.
Central to the perceptiveModels platform are the following concepts:
- Sensor: a sensor is an object that simply allows retrieving raw data from a source, e.g. a port, a motor interface (example: the patches of a finger tip are considered to be sensors).
- Node: a node is an object that performs some operations on the data that it can read through the sensors attached to the node itself (example: a finger can be treated as a node to which a number of patches are attached).
- Model: a model is a kind of super-object that encapsulates nodes and executes operations on these nodes when asked by the user; it also provides the results in a unified format, so that the user might instantiate different models to easily change the implementation without affecting the format of the outcome.
