Skip to content
RobotWorld
← Back to Videos

FACT: Learning from failure in robot learning

Loading video
Loading video

Children learn to walk by falling. FACT lets robots learn from failed rollouts, which reveal what bad actions look like—information that successful demonstrations alone cannot provide.

Embodied AIRobot LearningFailure LearningFACTReinforcement Learning
Category: research
Author: @QuanquanPeng03
Date: 2026-08-12T00:00:00
Duration: 136.746s
Source: https://x.com/QuanquanPeng03/status/2087581568533680308
Reference: https://t.co/XlbdeBcGCg

Related Videos

MessyMem: Persistent Spatial Memory for Robots
0:48

MessyMem: Persistent Spatial Memory for Robots

@anuvabanwasi · 856 views · 2026-09-16
Persistent MemoryMobile Manipulation3D scene graphs
Odyssey-3: A Foundation World Model for Physical Agents
1:37

Odyssey-3: A Foundation World Model for Physical Agents

@odysseyml · 1.2k views · 2026-09-15
World ModelsFoundation ModelsPhysical AI
VLA-Precision: one hour of real-robot online RL reaches 98.3% success
0:27

VLA-Precision: one hour of real-robot online RL reaches 98.3% success

@heetezition · 977 views · 2026-09-08
VLAReinforcement LearningRobot Learning
Same VLA, three data sources: Human Archive OTS wins on task success
0:43

Same VLA, three data sources: Human Archive OTS wins on task success

@babugi28 · 19 views · 2026-10-15
VLAPre-training DataAblation
Gen-HumanEgo: Open-Source Egocentric Human Data for Robot Learning
4:14

Gen-HumanEgo: Open-Source Egocentric Human Data for Robot Learning

@GenrobotAI · 970 views · 2026-09-23
Egocentric VisionRobot LearningOpen Source
AGIBOT Open-Sources Its World 2026 RL Dataset
0:28

AGIBOT Open-Sources Its World 2026 RL Dataset

@spaceandtech_ · 770 views · 2026-09-03
AgiBotEmbodied AIReinforcement Learning

As an Amazon Associate, we earn from qualifying purchases.