GACL: Grounded Adaptive Curriculum Learning with Active Task and Performance Monitoring

TL;DR

This paper introduces GACL, a novel curriculum learning framework for robotics that adaptively generates tasks based on performance and maintains relevance to target domains, improving success rates in navigation and locomotion tasks.

Contribution

GACL presents a new task representation, active performance monitoring, and grounding approach tailored for robotics curriculum learning, addressing limitations of manual and automated methods.

Findings

Achieved 6.8% higher success in wheeled navigation.

Achieved 6.1% higher success in quadruped locomotion.

Validated effectiveness across multiple robotics tasks.

Abstract

Curriculum learning has emerged as a promising approach for training complex robotics tasks, yet current applications predominantly rely on manually designed curricula, which demand significant engineering effort and can suffer from subjective and suboptimal human design choices. While automated curriculum learning has shown success in simple domains like grid worlds and games where task distributions can be easily specified, robotics tasks present unique challenges: they require handling complex task spaces while maintaining relevance to target domain distributions that are only partially known through limited samples. To this end, we propose Grounded Adaptive Curriculum Learning, a framework specifically designed for robotics curriculum learning with three key innovations: (1) a task representation that consistently handles complex robot task design, (2) an active performance tracking mechanism that allows adaptive curriculum generation appropriate for the robot's current capabilities, and (3) a grounding approach that maintains target domain relevance through alternating sampling between reference and synthetic tasks. We validate GACL on wheeled navigation in constrained environments and quadruped locomotion in challenging 3D confined spaces, achieving 6.8% and 6.1% higher success rates, respectively, than state-of-the-art methods in each domain.