Skip to content
Agientry
← Back to Videos

LeWAM: An End-to-End JEPA That Controls Real Robots

Loading video
Loading video

LeWAM is an end-to-end JEPA for control: joint next-latent prediction and action flow matching let one 17M-param model act directly and plan visually, lifting contact-rich success from 28.6% to 89.7% on a real robot.

World ModelsJEPAManipulationMotion PlanningRobot LearningRobot Manipulation
Category: arm
Author: @Minghao__Fu
Date: 2026-10-09T00:00:00
Duration: 57.25s
Source: https://x.com/Minghao__Fu/status/2108638867326968196
Reference: https://le-wam.github.io/

Related Videos

Proxy Policy Steering: Test-Time Robot Policy Adaptation
0:43

Proxy Policy Steering: Test-Time Robot Policy Adaptation

@LeoKharon · 547 views · 2026-09-26
Proxy Policy SteeringRobot Policy AdaptationDiffusion policies
Graph-as-Policy + Rust Executor: 5.42s vs 8.79s Median Pickup in YAM Sim
0:11

Graph-as-Policy + Rust Executor: 5.42s vs 8.79s Median Pickup in YAM Sim

@isthatdebbiej · 542 views · 2026-09-21
Graph PolicyRobot LearningManipulation
DexTacWAM: Making Touch Part of the Robot World Model
0:31

DexTacWAM: Making Touch Part of the Robot World Model

@robotsdigest · 508 views · 2026-09-22
Tactile SensingWorld ModelsBimanual Manipulation
NodeSLAM: Precise Stacking of Never-Seen Objects via Object-Model SLAM
0:58

NodeSLAM: Precise Stacking of Never-Seen Objects via Object-Model SLAM

@AjdDavison · 496 views · 2026-09-20
Object-Centric SLAMNodeSLAMMotion Planning
Flex-pi: A Multi-Stream World-Action Model with Compute Flexibility
0:43

Flex-pi: A Multi-Stream World-Action Model with Compute Flexibility

@LeoKharon · 846 views · 2026-09-08
VLAWorld ModelsRobot Manipulation
DemoMimic: One Demo, Many Unseen Objects
0:40

DemoMimic: One Demo, Many Unseen Objects

@heetezition · 1.1k views · 2026-09-03
DemoMimicDexterous ManipulationContact Geometry

As an Amazon Associate, we earn from qualifying purchases.