Skip to content

Changes

Page history
Update home authored by Serawork Amsalu Wallelign's avatar Serawork Amsalu Wallelign
Hide whitespace changes
Inline Side-by-side
home.md
View page @ 65689651
This is the documentation for the source code of GUI for LfD. The app is designed in [Android Studio](https://developer.android.com/studio/intro) and works only in android devices.
This is the documentation for the source code of GUI for Learn from Demonistration (LfD). The app is designed in [Android Studio](https://developer.android.com/studio/intro) and works only in android devices.
The aim of the project is to design a tool for end-users to program robots for a specific task.
Currently, robots are programmed by knowledge engineers to perform some application
which requires re-programming if the application or the environment changes. However,
with the increased popularity of robots, it is important to provide users with such
capabilities. In learning from demonstrations, robots autonomously learn knowledge model from observations of user's performing a task. This repository contains a graphical user interface for LfD using [LOCM](https://gricad-gitlab.univ-grenoble-alpes.fr/wallelis/locm) algorithm to learn domain. <br /><br />
capabilities. In learning from demonstrations, robots autonomously learn knowledge model from observations of user's performing a task. This repository contains a graphical user interface for LfD using [LOCM](https://gricad-gitlab.univ-grenoble-alpes.fr/wallelis/locm) algorithm to learn domain knowledge. <br /><br />
![lfd](uploads/ac5f343ae716aefd83e42259101d6b04/lfd.png)
It is an iterative process where the user demonistrates a task, learn a domain using LOCM, test the generated model and repeat the pocess guided by the output from LOCM to provide more meaningful input sequences. The figure below shows the flow diagram of the whole process.<br /><br />
![flow_diagram](uploads/469b78742ba765fce3b16846d3d8c27f/flow_diagram.png) <br /><br />
The user first performs demonistration of a task (creates sequence of valid actions). Every time a new action is added, the system checks if the sequence is enough to train LOCM and prompts the user. Once a valid sequence is created, a model (in PDDL) is generated using LOCM. The user can view, modify and test the domain. The process is repeated until a working domain is generated.
The GUI has the following functionalities: <br />
- [Create or open existing project](Create a Project)
......