Policy-Guided World Model Planning for Language-Conditioned Visual Navigation

TL;DR

PiJEPA is a two-stage framework combining learned policies and latent world model planning, significantly improving instruction-conditioned visual navigation in real-world tasks.

Contribution

The paper introduces PiJEPA, integrating a finetuned policy with a latent world model for faster, more accurate navigation based on natural language instructions.

Findings

PiJEPA outperforms standalone policies in goal-reaching accuracy.

Using policy-derived warm-start improves world model planning convergence.

Comparison of vision encoders DINOv2 and V-JEPA-2 shows impact on performance.

Abstract

Navigating to a visually specified goal given natural language instructions remains a fundamental challenge in embodied AI. Existing approaches either rely on reactive policies that struggle with long-horizon planning, or employ world models that suffer from poor action initialization in high-dimensional spaces. We present PiJEPA, a two-stage framework that combines the strengths of learned navigation policies with latent world model planning for instruction-conditioned visual navigation. In the first stage, we finetune an Octo-based generalist policy, augmented with a frozen pretrained vision encoder (DINOv2 or V-JEPA-2), on the CAST navigation dataset to produce an informed action distribution conditioned on the current observation and language instruction. In the second stage, we use this policy-derived distribution to warm-start Model Predictive Path Integral (MPPI) planning over a separately trained JEPA world model, which predicts future latent states in the embedding space of the same frozen encoder. By initializing the MPPI sampling distribution from the policy prior rather than from an uninformed Gaussian, our planner converges faster to high-quality action sequences that reach the goal. We systematically study the effect of the vision encoder backbone, comparing DINOv2 and V-JEPA-2, across both the policy and world model components. Experiments on real-world navigation tasks demonstrate that PiJEPA significantly outperforms both standalone policy execution and uninformed world model planning, achieving improved goal-reaching accuracy and instruction-following fidelity.