Skip to content
RobotWorld
← Back to Videos

RT-SAFE: real-time embodied-agent safety benchmarking

Loading video
Loading video

RT-SAFE evaluates eight vision-language models while the world keeps moving: task success reaches 94.1%, but safe completion is only 0.7%, with 12.3x more collisions than static evaluation.

Embodied AISafety EvaluationReal-Time ReasoningVision-language modelsAutonomous NavigationCollision AvoidanceBenchmarksRT-SAFEReal-Time Inference
Category: research
Author: @XTRose88
Date: 2026-10-06T00:00:00
Duration: 60.0s
Source: https://x.com/XTRose88/status/2107282503711203646
Reference: https://xtrose29.github.io/RT-SAFE/

Related Videos

MessyMem: Persistent Spatial Memory for Robots
0:48

MessyMem: Persistent Spatial Memory for Robots

@anuvabanwasi · 620 views · 2026-09-16
Persistent MemoryMobile Manipulation3D scene graphs
Show-Harness: Letting a VLM Agent Play Real Robots
1:50

Show-Harness: Letting a VLM Agent Play Real Robots

@ZechenBai · 652 views · 2026-09-10
VLAVLM AgentRobot Manipulation
Reka Releases Rho-1, a 19B Omni Model
1:00

Reka Releases Rho-1, a 19B Omni Model

@RekaAILabs · 24 views · 2026-10-05
Embodied AIOmni ModelMultimodal
Awesome-UMI Adds 9 New Device Projects
0:28

Awesome-UMI Adds 9 New Device Projects

@ArataHorie · 648 views · 2026-09-26
RoboticsImitation LearningData Collection
Gen-HumanEgo: Open-Source Egocentric Human Data for Robot Learning
4:14

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

@GenrobotAI · 705 views · 2026-09-23
Egocentric VisionRobot LearningOpen Source
Odyssey-3: World Models as Training Grounds for Agents
1:28

Odyssey-3: World Models as Training Grounds for Agents

@olivercameron · 586 views · 2026-09-15
World ModelsEmbodied AISimulation

As an Amazon Associate, we earn from qualifying purchases.