# A learned goal metric lifts planning success in four benchmarks

> A new arXiv paper reports that ranking states by time to a goal raised average planning success from 32.7% to 63.0% across four benchmarks.

Oossa · 2026-11-05 · https://oossa.com/en/a-learned-goal-metric-lifts-planning-success-in-four-benchmarks

Event date: 2026-11-04

A new method called Time-to-Goal (T2G) aims to help robots plan longer routes by giving lower costs to states that are closer in time to a goal. The researchers train it on offline trajectory data, without reward or policy labels, and use it in place of the usual Euclidean goal cost while keeping the world-model encoder and predictor frozen.

In results reported in the paper, T2G raised average planning success from 32.7% to 63.0% across four navigation and manipulation benchmarks, approaching or sometimes exceeding hierarchical JEPA baselines. The paper was submitted to arXiv on November 4, 2026.

## The facts

- Reported average planning success: 32.7% with flat LeWM; 63.0% with T2G across four benchmarks.
- The paper says T2G learns from offline trajectories and requires no reward or policy labels.

## Why it matters

For robotics researchers using flat world models, the reported results suggest a learned goal-cost metric could improve benchmark planning success without retraining the model's encoder or predictor; performance on real robots is not established here.

## Sources & references

1. [Time-to-Goal: Learned Temporal Metrics for Long-Horizon Planning with JEPA World Models](https://arxiv.org/abs/2611.05847) – arXiv cs.RO, 2026-11-05

Last updated: 2026-11-05