Final Post

This is the final post in the series of posts describing the work done in GSoC.

Work Done

In the final coding period, I wrote documentation explaining how one can use this project and how to use pytorch and the two frameworks(dgl and pytorch geometric) to develop machine learning models to apply on a graph dataset. I also developed an API for predicting scores for Social Scenarios. Finally, after testing the code and adding documentation I got the code merged into the a Robocomp repository which is linked below.

Final Work Product

SNGNN: Github Repository

• Pull Request : Link

Graph Neural Network Models: Github Repository Link

Robocomp Agents: Github Repository Link

Social Navigation Scoring API: Github Repository Link

Screenshots

| Scenario 1 | Scenario 2 | Scenario 3| |— | — | — | | | | |

Future Scope

1. Develop manual editor for the social scenario
2. Add more functionality to the Social Navigation API

Learnings

This experience has given me immense learnings:

1. Write clean code and develop solutions which are complete without using any work-arounds
2. Using git effectively
3. Writing documentation
4. Designing good structured code which is extensible
5. Work in a collaborative environment which required soft skills
6. Develop machine learning models in pytorch

Finally I would like to thank my mentor, Luis for all the help and guidance.

Ronit Jorvekar