#Robot Learning (39)

Waterfall Pagination

1:04

GE-Act 2.0: Pretraining and Scaling a World-Action Model for Robotic Manipulation

GE-Act 2.0 combines a control-oriented autoencoder, a single-step visual planner, an inverse dynamics model, and knowledge-aligned selective optimization to pretrain a world-action policy from scratch on manipulation data. Scaling co-training data from 300 to 30,000 hours raises zero-shot OOD success to 44.1% on G1-OP and 31.1% on G2-90D without task-specific fine-tuning.

AgiBot Research Team, Renhang Liu, Wenzhi ZhaoSep 4, 2026

World-action modelsRobot ManipulationEmbodied AISep 4, 2026

3:30

microSLAM: Monocular SLAM for Interactive 3D Environments

@ZenoInMotion · 175 views · 2026-09-10

SLAMMonocular Vision3D reconstruction

0:25

A Wrist Camera Cuts the Cost of Robot Data Collection

@CyberRobooo · 174 views · 2026-09-10

Robot LearningData CollectionWearable Camera

0:59

StereoPolicy Learns 3D Structure Directly From Stereo Images

@LambdaAPI · 150 views · 2026-09-09

Stereo visionRobot Learning3D Perception

3:00

OpenWAM open world-action model pretraining framework

@KnightNemo_ · 166 views · 2026-09-09

World ModelsRobot LearningPretraining

1:14

Genrobot vision-only full-body motion capture

@XRoboHub · 132 views · 2026-09-10

Robot LearningMotion CaptureVision

2:20

GRID Auto-Engineering: Automating How Robot Intelligence Gets Built and Deployed

@genrobotics_ai · 107 views · 2026-09-09

Robotics PlatformAuto-EngineeringDeployment

0:27

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

@heetezition · 440 views · 2026-09-08

VLAReinforcement LearningRobot Learning

0:17

GPT-6 Runs Real2Sim End to End from Robot Demonstrations

@Lingxiao234 · 861 views · 2026-09-07

Real-to-SimGPT-6Robot Learning

1:15

Figure Stacks Both Halves of the Physical AI Bottleneck

@adrianmacneil · 198 views · 2026-09-07

Figure AIPhysical AIRobot Data

0:32

TAM: reusable torque adaptation for manipulation

@dongwonson110 · 310 views · 2026-09-05

TAMTorque AdaptationSim-to-Real

0:17

VLAct: representation-centric VLA pre-training

@yokoshaan · 317 views · 2026-09-05

VLActVLACross-Embodiment

2:06

RoboTok: Human Demonstration Data Engine

@xiynfi1520580 · 395 views · 2026-09-04

RoboTokRICENVIDIA

Efficient On-policy Visual-RL via Stochastic Decoupled Policy Gradient

We present the stochastic decoupled policy gradient (SDPG), a lightweight visual reinforcement learning (RL) method that trains diverse visuomotor control policies end-to-end within a few hours on a single NVIDIA RTX 4080 GPU. SDPG estimates policy gradients via random perturbations of trajectory rollouts, requiring orders of magnitude fewer batch-rendered environments and substantially reducing compute and memory overhead. On visual MuJoCo benchmarks, SDPG consistently outperforms baseline methods in training time, memory usage, and rewards. Finally, to support future research, we introduce a suite of realistic visual robotics benchmarks spanning dexterous manipulation, challenging locomotion, and demonstrate effective sim-to-real transfer on physical hardware.

Haoxiang You, Yilang Liu, Davis ZongMay 26, 2026

Sim-to-RealVisual reinforcement learningRobot LearningMay 26, 2026