Skip to content
RobotWorld
← Back to Videos

ACT-2 Preview — first robotics model unifying generalization with reliability

Loading video
Loading video

Tony Zhao's ACT-2 Preview: a single fine-tuning example teaches a new generalizable behavior — zero-shot, real unseen homes, 99% success rate.

VLAACT-2HumanoidEmbodied AIGeneralizationEnd-to-EndHome Service
Category: perception
Author: @tonyzzhao
Date: 2026-07-16T00:00:00
Duration: 145.451s
Source: https://x.com/tonyzzhao/status/2077806003308179802
Reference: https://tonyzzhao.com/

Related Videos

Astribot Lumo-2 — 20+ household tasks, one foundation model, real robot
4:57

Astribot Lumo-2 — 20+ household tasks, one foundation model, real robot

@XRoboHub · 485 views · 2026-07-16
VLAAstribotLumo-2
Early access to NVIDIA Isaac GR00T N1.7 is here 🎉 — an open, commerci...
0:10

Early access to NVIDIA Isaac GR00T N1.7 is here 🎉 — an open, commerci...

@NVIDIARobotics · 513 views · 2026-04-17
NVIDIA GR00THumanoidIsaac
MiniCPM-Robot open-sourced — 1.5B general-purpose VLA model
1:17

MiniCPM-Robot open-sourced — 1.5B general-purpose VLA model

@OpenBMB · 430 views · 2026-07-19
VLAOpenBMBMiniCPM
Grounded API: Robot Perception and SLAM
1:14

Grounded API: Robot Perception and SLAM

@GroundedSI · 766 views · 2026-09-16
Robot perceptionSLAMHand Tracking
DeepMind D4RT turns a video into a 4D world
3:33

DeepMind D4RT turns a video into a 4D world

@Alacritic_Super · 808 views · 2026-08-31
D4RT4D WorldWorld Models
LiteReality-Agent turns room scans into interactive 3D scenes
0:58

LiteReality-Agent turns room scans into interactive 3D scenes

@ZheningHuang · 453 views · 2026-08-07
LiteRealityRGB-DScene Reconstruction

As an Amazon Associate, we earn from qualifying purchases.