Search papers, labs, and topics across Lattice.
This paper introduces Attacca, a novel framework for training visual goal-conditioned policies for long-horizon embodied agents operating under state continuity. By decoupling goal images from execution environments and employing context-decoupled goal sampling, Attacca enables agents to ground targets even when they are not directly visible. Evaluation in Minecraft demonstrates significant improvements in both short- and long-horizon task completion, outperforming baselines by up to 7x.
Embodied agents can now reliably achieve complex, multi-stage goals in dynamic environments by learning to predict and ground targets from decoupled visual context, even when the goal is out of view.
A central capability of embodied agents is to accomplish complex objectives through sequences of interdependent tasks. Yet existing visual goal-conditioned policies underlying these agents are typically evaluated on isolated interactions where the target is already visible, and thus do not capture the conditions that arise during continuous long-horizon task execution. In such settings, each task begins from the state left by the previous one: the agent may end at a different position and orientation, the world may have been modified, and the next interaction target may lie outside the current field of view. As a result, agents relying on such policies may struggle to proceed to the next task when they cannot ground their target in the current observation. To address this challenge, we propose Attacca, a new approach that trains visual goal-conditioned policies on complete search-to-interact trajectories using goal images decoupled from the execution environment. Attacca uses context-decoupled goal sampling to pair each demonstration with a class-compatible masked goal image from another world, removing direct scene and pose correspondence. It learns dense current-view grounding through a target-mask prediction head, providing auxiliary supervision beyond action imitation. We further introduce behavioral-phase conditioning that teaches the policy to distinguish Search, Approach, and Interact stages and adapt its control as execution progresses. We evaluate Attacca on multiple short- and long-horizon embodied tasks in Minecraft. Our method achieves 39.0-47.5% clean success, improving over the strongest baseline by 1.7-2.4x. On long-horizon tasks, it attains 54%, 30%, and 28% completion, yielding up to a 7x improvement.
