Edinburgh Research Explorer

From explanation to synthesis: Compositional program induction for learning from demonstration

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Original languageEnglish
Title of host publicationRobotics: Science and System XV
Subtitle of host publicationRSS XV
EditorsAntonio Bicchi, Hadas Kress-Gazit, Seth Hutchinson
Place of PublicationFreiburg im Breisgau, Germany
Number of pages10
ISBN (Electronic)978-0-9923747-5-4
Publication statusE-pub ahead of print - 26 Jun 2019
EventRobotics: Science and Systems 2019 - Freiburg im Breisgau, Germany
Duration: 22 Jun 201926 Jun 2019


ConferenceRobotics: Science and Systems 2019
Abbreviated titleRSS 2019
CityFreiburg im Breisgau
Internet address


Hybrid systems are a compact and natural mechanism with which to address problems in robotics. This work introduces an approach to learning hybrid systems from demonstrations, with an emphasis on extracting models that are explicitly verifiable and easily interpreted by robot operators. We fit a sequence of controllers using sequential importance sampling under a generative switching proportional controller task model. Here, we parameterise controllers using a proportional gain and a visually verifiable joint angle goal. Inference under this model is challenging, but we address this by introducing an attribution prior extracted from a neural end-to-end visuomotor control model. Given the sequence of controllers comprising a task, we simplify the trace using grammar parsing strategies, taking advantage of the sequence compositionality, before grounding the controllers by training perception networks to predict goals given images. Using this approach, we are successfully able to induce a program for a visuomotor reaching task involving loops and conditionals from a single demonstration and a neural endto-end model. In addition, we are able to discover the program used for a tower building task. We argue that computer programlike control systems are more interpretable than alternative endto-end learning approaches, and that hybrid systems inherently allow for better generalisation across task configurations.


Robotics: Science and Systems 2019


Freiburg im Breisgau, Germany

Event: Conference

Download statistics

No data available

ID: 87641567