Timezone: »

Air Learning: An End To End Learning Gym For Aerial Robots
Srivatsan Krishnan · Colby Banbury · Bardienus Duisterhof · Aleksandra Faust · Vijay Janapa Reddi

Mon Mar 02 04:00 PM -- 07:00 PM (PST) @ Ballroom B + C #104

Air Learning is an interdisciplinary, open-source research infrastructure that aims to soften the boundaries between aerial robotics, machine learning, controls, and systems architecture. It provides the tooling and various components necessary for developing an end-to-end learning-based application for aerial robotics starting from simulation to deployment on a real aerial robot. By having all the key components tightly integrated, we hope researchers can use this tool to develop novel solutions to several open problems in these domains. We also hope researchers can use it to explore and understand various tradeoffs of their solution due to cross-domain interactions between algorithms and systems. Also since the infrastructure is opensource, the research community can add new features thus modifying it according to their own requirements and use cases.

Author Information

Srivatsan Krishnan (Harvard University)
Colby Banbury (Harvard)
Bardienus Duisterhof (Harvard University)
Aleksandra Faust (Google)
Vijay Janapa Reddi (Harvard University)

More from the Same Authors