Skip to content
RobotWorld
← Back to Videos

DISCO Distills Continuous-Action Tree Search into Policies

Loading video
Loading video

DISCO runs GPU-parallel MCTS with a learned policy proposing actions and a value function scoring them, then distills the search back into a fast policy that deploys without a simulator or test-time search.

VLARobot ManipulationTree SearchMCTSPolicy learning
Category: research
Author: @robotsdigest
Date: 2026-11-04T00:00:00
Duration: 35.666s
Source: https://x.com/robotsdigest/status/2118042791137935705

Related Videos

GPT-6 Astra Tops Independent Robot Manipulation Eval at 70.5%
0:07

GPT-6 Astra Tops Independent Robot Manipulation Eval at 70.5%

@yokoshaan · 666 views · 2026-09-21
VLARobot Manipulation
RD-VLA Scales Robot Reasoning in Latent Space
1:23

RD-VLA Scales Robot Reasoning in Latent Space

@DJiafei · 754 views · 2026-02-10
VLALatent Reasoning
Awesome-UMI Adds 9 New Device Projects
0:28

Awesome-UMI Adds 9 New Device Projects

@ArataHorie · 988 views · 2026-09-26
RoboticsImitation Learning
Real-Time EXPO-FT: Fast, Reliable RL for Real-Time VLA Policies
1:00

Real-Time EXPO-FT: Fast, Reliable RL for Real-Time VLA Policies

@perryadong · 597 views · 2026-09-21
VLAReinforcement Learning
Opus 5 fails a simulated toxic-lab safety task
0:39

Opus 5 fails a simulated toxic-lab safety task

@ariansadeg · 778 views · 2026-09-19
Physical AI SafetyVLA
Show-Harness: Letting a VLM Agent Play Real Robots
1:50

Show-Harness: Letting a VLM Agent Play Real Robots

@ZechenBai · 1.1k views · 2026-09-10
VLAVLM Agent

As an Amazon Associate, we earn from qualifying purchases.