Experimental platform for Science

The Poppy platform was initially designed within a publicly funded scientific project (Flowers project-team funded by ERC Grant Explorers, Inria and Région Aquitaine). It targeted to respond jointly to three complementary needs in robotics and cognitive science.

Make the body an experimental variable

The properties of the body (shapes, elasticity, distribution of mass…) have a crucial impact on sensorimotor control, cognitive skills and social interaction. For example, adequate shapes of legs can considerably simplify the acquisition of locomotion. The social and emotional response of a human in front of a robot is also strongly impacted by its body appearance. Thus, a scientific enquiry of the role of the body requires the possibility of fast design, building and experimentation of alternative morphologies. 3D printing and other rapid prototyping techniques now make it possible, and this was leveraged in the design of the Poppy platform.

Make scientific output openly accessible, reproducible and cumulative

Robots are often complex systems, and this has so far often resulted in robotics research project which results were relying on either closed platforms, hiding sometimes crucial details, or experiments which cannot be reproduced.

This is a barrier to scientific progress. The Poppy platform is a step towards addressing this issue, being accessible both in terms of cost and complexity, and allowing researchers to share hardware and experimental details in addition to code and algorithms.

Make scientific output usable outside science labs

A number of research projects in science labs could in principle be reused in educational, FabLab or industrial projects. Yet, the difficulty of accessing information through the formalism of scientific publications is often a strong obstacle.

The sharing of experimental materials through the Poppy platform opens new possibilities in this perspective.

How to use Poppy for scientific work

TEDx talk discussing how Poppy opens scientific possibilities to explore the mechanisms of learning and development with open source baby robots.

The Poppy platform allows users to create and modify easily the shape of robots. Poppy creatures can also be used directly as reproducible and accessible experimental platforms.

Investigate quickly, and at a reasonable cost, various mechanical design for body parts such as the leg, or test several mechanism for the feet, and study their impact on balance or energy consumption.

Explore how different body appearances can impact human-robot interaction; for instance, how humans perceive the movements or the internal state of the robot.

Test how autonomous learning algorithms allow a robot to learn how to control new morphologies and interact with objects, such as with the compatible open-source Explauto python library for curiosity-driven learning.

Thanks to the availability of simulators, one can also study how skills acquired by the robot in simulation (e.g. through stochastic optimization) can be transferred efficiently to the physical robot.

Poppy being accessible, other science lab in the world will be able to reproduce your research and extend it with their own idea. Thanks to the 3D printing technology, off the shelf affordable components, and true open source community, you can have very short cycles of development. Your research can go fast.

How can I contribute?

  • By investigating your own research question using Poppy and sharing the source code of the software and hardware to the community such that anyone can reproduce you results and test on their own platform;

  • By creating tutorials allowing other researchers to reproduce your experiments;

  • By connecting your latest machine learning or control algorithms to the pypot library so that users in the Poppy community can reuse them for their projects;

  • By submitting scientific challenges to the Poppy community;

  • By helping others to solve their scientific goals.