Autonomous skill discovery with Quality-Diversity and Unsupervised Descriptors
This addresses the need for more versatile and resilient robots by automating descriptor definition, though it is incremental as it builds on existing Quality-Diversity and unsupervised methods.
The paper tackles the problem of manually defining behavioral descriptors in Quality-Diversity optimization for robotics by introducing a method that combines it with unsupervised dimensionality reduction, enabling robots to autonomously discover a large range of behaviors without prior knowledge, with results showing behaviors similar to handcrafted solutions and more diverse than existing unsupervised methods.
Quality-Diversity optimization is a new family of optimization algorithms that, instead of searching for a single optimal solution to solving a task, searches for a large collection of solutions that all solve the task in a different way. This approach is particularly promising for learning behavioral repertoires in robotics, as such a diversity of behaviors enables robots to be more versatile and resilient. However, these algorithms require the user to manually define behavioral descriptors, which is used to determine whether two solutions are different or similar. The choice of a behavioral descriptor is crucial, as it completely changes the solution types that the algorithm derives. In this paper, we introduce a new method to automatically define this descriptor by combining Quality-Diversity algorithms with unsupervised dimensionality reduction algorithms. This approach enables robots to autonomously discover the range of their capabilities while interacting with their environment. The results from two experimental scenarios demonstrate that robot can autonomously discover a large range of possible behaviors, without any prior knowledge about their morphology and environment. Furthermore, these behaviors are deemed to be similar to handcrafted solutions that uses domain knowledge and significantly more diverse than when using existing unsupervised methods.