Zero-Splat TeleAssist: A Zero-Shot Pose Estimation Framework for Semantic Teleoperation

ICRAW 25 2025 30.1 system, application

TLDR

A zero-shot sensor-fusion pipeline using CCTV, segmentation, depth, PCA, and 3DGS for 6-DoF pose estimation in multilateral teleoperation.

Reasoning

The paper presents a novel integration of vision-language and 3DGS for real-time pose estimation without fiducials, which is a strength. However, it lacks explicit evaluation on standard benchmarks or comparison to existing methods, and the claimed 'world model' is narrowly scoped to teleoperation, not general world simulation or dynamics prediction.

Read-first score

Read-first score 30.1, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 9.

Recency 8%

86.7

Uses a gentle age decay so recent papers surface without erasing older foundations. 2025

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=none

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Topical relevance 42%

12.9

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 88.

Keyword Scores

world model

world simulator

generative world model

interactive world model

video world model

world dynamics prediction

model-based reinforcement learning world model

Deep Analysis

Innovations

Zero-shot sensor-fusion pipeline combining vision-language segmentation, monocular depth, weighted-PCA pose extraction, and 3D Gaussian Splatting for teleoperation
Transforms commodity CCTV streams into a shared 6-DoF world model without fiducials or depth sensors
Enables multilateral teleoperation with real-time global positions and orientations of multiple robots

Methodology

The pipeline integrates vision-language segmentation to identify robots, monocular depth estimation to obtain 3D information, weighted-PCA for extracting 6-DoF poses, and 3D Gaussian Splatting to build a shared world model from CCTV streams. This zero-shot approach requires no prior training or calibration for new environments.

Key Results

The framework provides real-time global positions and orientations of multiple robots in a multilateral teleoperation setup, operating without fiducial markers or dedicated depth sensors.

Resources

Paper