ARTPS: Depth-Enhanced Hybrid Anomaly Detection and Learnable Curiosity Score for Autonomous Rover Target Prioritization

We present ARTPS (Autonomous Rover Target Prioritization System), a novel hybrid AI system that combines depth estimation, anomaly detection, and learnable curiosity scoring for autonomous exploration of planetary surfaces. Our approach integrates monocular depth estimation using Vision Transformers with multi-component anomaly detection and a weighted curiosity score that balances known value, anomaly signals, depth variance, and surface roughness.

The system achieves significant performance improvements with AUROC of 0.894, AUPRC of 0.847, and F1-Score of 0.823 on Mars rover datasets. We demonstrate enhanced target prioritization accuracy through ablation studies and provide comprehensive analysis of component contributions. The hybrid fusion approach reduces false positive rate to 0.089 while maintaining high detection sensitivity across diverse terrain types.

Key contributions include: (1) Integration of single-image depth estimation with anomaly detection, (2) Multi-component fusion strategy combining image and depth cues, (3) Learnable curiosity score balancing novelty and known value, (4) Comprehensive evaluation on Mars rover datasets, and (5) Real-time performance optimization for edge computing constraints.

This work advances autonomous scientific exploration by providing explainable target prioritization suitable for operator-in-the-loop workflows in planetary exploration missions.