TP-MDDN: Task-Preferenced Multi-Demand-Driven Navigation with Autonomous Decision-Making

In daily life, people often move through spaces to find objects that meet their needs, posing a key challenge in embodied AI. Traditional Demand-Driven Navigation (DDN) handles one need at a time but does not reflect the complexity of realworld tasks involving multiple needs and personal choices. To bridge this gap, we introduce Task-Preferenced Multi-Demand-Driven Navigation (TP-MDDN), a new benchmark for long-horizon navigation involving multiple sub-demands with explicit task preferences. To solve TP-MDDN, we propose AWMSystem, an autonomous decision-making system composed of three key modules: Break-LLM (instruction decomposition), LocateLLM (goal selection), and StatusMLLM (task monitoring). For spatial memory, we design MASMap, which combines 3D point cloud accumulation with 2D semantic mapping for accurate and efficient environmental understanding. Our Dual-Tempo action generation framework integrates zero-shot planning with policy-based fine control, and is further supported by an Adaptive Error Corrector that handles failure cases in real time. Experiments demonstrate that our approach outperforms state-of-the-art baselines in both perception accuracy and navigation robustness.

Recent work like MO-DDN [5] addresses navigation instructions through Multi-object Demanddriven Navigation with human preferences. For example, given an instruction like "I need to display my photography collection, preferably with good lighting", the demand may involve multiple objects such as picture frames, bookshelves, and ceiling lamps. However, it still focuses on singledemand navigation. Some works have explored long-range instruction navigation by constructing Landmark Semantic Memory for decision-making planning [6] or adopting Autonomous Evolution mechanisms [7,8], but they primarily target object-driven navigation.