Gradient-based Planning for World Models at Longer Horizons
Lab/vendor primary source
#213
- Canonical URL
- http://bair.berkeley.edu/blog/2026/04/20/grasp/
- Topic
- Practitioner Notes
- First seen
- 2026-07-16 19:08:00
- Last seen
- 2026-07-16 19:08:00
Source raw items (1)
- Blog / Newsletter2026-07-16 19:07:20Gradient-based Planning for World Models at Longer Horizons
<!-- twitter --> <div style="display: flex; margin-bottom: 1.5em;"> <img alt="BallNav demo" src="https://bair.berkeley.edu/static/blog/grasp/ballnav_demo.gif" /> <img alt="Push-T demo" src="https://bair.berkeley.edu/static/blog/grasp/pusht_zoomout.gif" /> </div> <p><strong>GRASP</strong> is a new gradient-based planner for learned dynamics (a “world model”) that makes long-horizon planning practical by (1) lifting the trajectory into virtual states so optimization is parallel across time, (2) adding stochasticity directly to the state iterates for exploration, and (3) reshaping gradients so actions get clean signals while we avoid brittle “state-input” gradients through high-dimensional vision models.</p> <!--more--> <p>Large, learned world models are becoming increasingly capable. They can predict long sequences of future observations in high-dimensional visual spaces and generalize across tasks in ways that were difficult to imagine a few years ago. As these mode