Automated Curriculum Learning for Embodied Agents: A Neuroevolutionary Approach

TL;DR

This paper introduces an evolutionary algorithm enhanced with automatic curriculum learning that adaptively selects environmental conditions to improve the robustness and performance of embodied agents without needing domain knowledge or extra hyperparameters.

Contribution

It presents a novel neuroevolutionary approach that automatically adjusts environmental difficulty, outperforming traditional methods on benchmark problems without requiring domain expertise.

Findings

Outperforms conventional algorithms on benchmark tasks

Generates solutions robust to environmental variations

Does not require domain knowledge or additional hyperparameters

Abstract

We demonstrate how an evolutionary algorithm can be extended with a curriculum learning process that selects automatically the environmental conditions in which the evolving agents are evaluated. The environmental conditions are selected so to adjust the level of difficulty to the ability level of the current evolving agents and so to challenge the weaknesses of the evolving agents. The method does not require domain knowledge and does not introduce additional hyperparameters. The results collected on two benchmark problems, that require to solve a task in significantly varying environmental conditions, demonstrate that the method proposed outperforms conventional algorithms and generates solutions that are robust to variations