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

2025-08-04

AI Summary

Current curriculum learning approaches for complex robotic tasks suffer from heavy reliance on manual design, poor generalizability, and low sample efficiency. To address these limitations in low-data regimes, this paper proposes an automated curriculum generation framework tailored for few-shot robotic learning. The method introduces a unified task embedding space, an active performance tracking mechanism, and a grounded sampling strategy that alternates between reference and synthetically generated tasks—enabling adaptive difficulty adjustment and dynamic preservation of target-domain distribution alignment. Integrating active learning, adaptive curriculum scheduling, and real-time performance evaluation, the framework is jointly optimized end-to-end with reinforcement learning. Evaluated on wheeled navigation and quadrupedal locomotion—both high-dimensional, sparse-reward robotic tasks—the approach achieves +6.8% and +6.1% success rate improvements over state-of-the-art methods, respectively, demonstrating superior effectiveness and robustness in data-constrained settings.

Technology Category

• Intelligent Robots: Learning & Optimization for ROB
• Machine Learning: Adversarial Learning & Robustness
• Planning, Routing, and Scheduling: Learning for Planning and Scheduling

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

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.

Problem

• Automates curriculum learning for complex robotics tasks
• Handles unknown target domain distributions with limited samples
• Improves success rates in navigation and locomotion tasks

Innovation

• Consistent task representation for complex robotics
• Active performance tracking for adaptive curriculum
• Grounding approach maintains target domain relevance