Learning What Matters: Adaptive Information-Theoretic Objectives for Robot Exploration

arXiv:2605.12084v1 Announce Type: cross
Abstract: Designing learnable information-theoretic objectives for robot exploration remains challenging. Such objectives aim to guide exploration toward data that reduces uncertainty in model parameters, yet it is often unclear what information the collected data can actually reveal. Although reinforcement learning (RL) can optimize a given objective, constructing objectives that reflect parametric learnability is difficult in high-dimensional robotic systems. Many parameter directions are weakly observable or unidentifiable, and even when identifiable directions are selected, omitted directions can still influence exploration and distort information measures. To address this challenge, we propose Quasi-Optimal Experimental Design (Qfootnotesize OED), an adaptive information objective grounded in optimal experimental design. Qfootnotesize OED (i) performs eigenspace analysis of the Fisher information matrix to identify an observable subspace and select identifiable parameter directions, and (ii) modifies the exploration objective to emphasize these directions while suppressing nuisance effects from non-critical parameters. Under bounded nuisance influence and limited coupling between critical and nuisance directions, Qfootnotesize OED provides a constant-factor approximation to the ideal information objective that explores all parameters. We evaluate Qfootnotesize OED on simulated and real-world navigation and manipulation tasks, where identifiable-direction selection and nuisance suppression yield performance improvements of SI35.23percent and SI21.98percent, respectively. When integrated as an exploration objective in model-based policy optimization, Qfootnotesize OED further improves policy performance over established RL baselines.