Complete code for implementing our socially guided continual learning system (SGCL): Integration of continual learning models with an android tablet and a mobile manipulator robt using ROS.
- ROS with a real Fetch robot or a simulator
- Android tablet
- android studio
- torch
- Scipy
- Scikit Learn
- Create
learned_objects,learned_models,test_objectsandcur_img_dirfolders. - Run
lifelong_model_server.pyfromlifelong_learningfolder. - Connect tablet with the server. Run
MainActivity.javafrom/home/sirrlab1/cl_hri/app/src/main/java/com/example/fetchgui_learning_testingin android studio to load the GUI on the tablet. - Run
fsilGUI.pyto connect the GUI with the server and the robot. - Use the GUI to continually teach and test the robot.
@inproceedings{
ayub2023clhri,
title={Human Perceptions of Task Load and Trust when Interactively Teaching a Continual Learning Robot},
author={Ali Ayub and Zachary De Francesco and Patrick Holthaus and Chrystopher L. Nehaniv and Kerstin Dautenhahn},
booktitle={IEEE/CVF CVPR 2023 (4th Workshop on Continual Learning in Computer Vision)},
year={2023}
}
@inproceedings{
ayub2023roman_continual,
title={How Do Human Users Teach a Continual Learning Robot in Repeated Interactions?},
author={Ali Ayub and Jainish Mehta and Zachary De Francesco and Patrick Holthaus and Kerstin Dautenhahn and Chrystopher L. Nehaniv},
booktitle={IEEE International Conference on Robot and Human Interactive Communication (ROMAN)},
year={2023}
}
@misc{ayub2023continual,
title={Continual Learning through Human-Robot Interaction -- Human Perceptions of a Continual Learning Robot in Repeated Interactions},
author={Ali Ayub and Zachary De Francesco and Patrick Holthaus and Chrystopher L. Nehaniv and Kerstin Dautenhahn},
year={2023},
eprint={2305.16332},
archivePrefix={arXiv},
primaryClass={cs.RO}
}