Skip to content
Agientry
← Back to Videos

FAR: Robot Arms Learn to Recover From Failure

Loading video
Loading video

FAR lets robot arms retry after failure without human corrections or environment resets. It reports 17.6% average success-rate gain in simulation and 11.7% in the real world.

Robot LearningManipulationFailure RecoveryTest-Time LearningFARCarnegie Mellon UniversityDiffusion policiesRobotic Arm
Category: arm
Author: @hrhao1
Date: 2026-10-26T00:00:00
Duration: 24.102s
Source: https://x.com/hrhao1/status/2114829486097367441
Reference: https://arxiv.org/abs/2607.01111

Related Videos

R2-WAM: data-free continual learning for world action models via rectified generative replay
1:14

R2-WAM: data-free continual learning for world action models via rectified generative replay

@Yunsong_Zhou · 103 views · 2026-10-21
ContinualLearningWorldActionModel
AlphaNIST: Binary Reward Plus Control Cost Hits 98.2% Sim Success
1:33

AlphaNIST: Binary Reward Plus Control Cost Hits 98.2% Sim Success

@octi_zhang · 182 views · 2026-10-13
Robotic ArmManipulation
AlphaNIST: One Policy From Scratch Solves All 20 NIST Assembly Tasks
0:07

AlphaNIST: One Policy From Scratch Solves All 20 NIST Assembly Tasks

@patrickhyin · 147 views · 2026-10-13
Robotic ArmManipulation
LeWAM: An End-to-End JEPA That Controls Real Robots
0:57

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

@Minghao__Fu · 242 views · 2026-10-09
World ModelsJEPA
Proxy Policy Steering: Test-Time Robot Policy Adaptation
0:43

Proxy Policy Steering: Test-Time Robot Policy Adaptation

@LeoKharon · 621 views · 2026-09-26
Proxy Policy SteeringRobot Policy Adaptation
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 · 621 views · 2026-09-21
Graph PolicyRobot Learning

As an Amazon Associate, we earn from qualifying purchases.