Hydra-Nav: Object Navigation via Adaptive Dual-Process Reasoning

While large vision-language models (VLMs) show promise for object goal navigation, current methods still struggle with low success rates and inefficient localization of unseen objects—failures primarily attributed to weak temporal-spatial reasoning. Meanwhile, recent attempts to inject reasoning into VLM-based agents improve success rates but incur substantial computational overhead. To address both the ineffectiveness and inefficiency of existing approaches, we introduce Hydra-Nav, a unified VLM architecture that adaptively switches between a deliberative "slow system" for analyzing exploration history and formulating high-level plans, and a reactive "fast system" for efficient execution. We train Hydra-Nav through a three-stage curriculum: (i) spatial-action alignment to strengthen trajectory planning, (ii) memory-reasoning integration to enhance temporal-spatial reasoning over long-horizon exploration, and (iii) iterative rejection fine-tuning to enable selective reasoning at critical decision points. Extensive experiments demonstrate that Hydra-Nav achieves state-of-the-art performance on the HM3D, MP3D, and OVON benchmarks, outperforming the second-best methods by 11.1\%, 17.4\%, and 21.2\%, respectively. Furthermore, we introduce SOT (Success weighted by Operation Time), a new metric to measure search efficiency across VLMs with varying reasoning intensity. Results show that adaptive reasoning significantly enhances search efficiency over fixed-frequency baselines. Project page is available at: https://zixuan-wang99.github.io/Hydra-Nav/.

Imagine asking a home robot to find an everyday object, like a TV. To succeed, the robot needs to explore its surroundings, remember which rooms it has already checked, and make logical guesses about where the object might be. Current robotic systems often struggle with this task: they either get lost easily because they lack a good understanding of their environment, or they stop to over-analyze every single step, which makes them incredibly slow. To solve this, we introduce HYDRA-NAV, a new robotic "brain" inspired by how humans naturally think. It relies on two distinct modes: A "fast" system: Used for simple, routine movements, such as walking straight down a clear hallway. A "slow" system: Used for deliberate, complex planning, such as reviewing past memories to decide which unexplored room to check next. Instead of constantly overthinking, our robot learns to automatically switch between these two modes, triggering the "slow" deep-thinking process only when it gets stuck, wanders in circles, or stops making progress. Our experiments demonstrate that this adaptive approach allows robots to locate objects much more successfully in entirely new, unseen homes, while significantly reducing the time it takes to complete the search.