LIVIN: Benchmarking Spatial and Embodied Intelligence in Digital Twins of Lived-In Homes

2026-10-08

AI Summary

This study addresses the challenge that existing embodied AI resources struggle to simultaneously achieve scale, realism, and interactivity, lacking faithful simulation of real residential layouts and physical constraints. To this end, we propose a digital twin construction paradigm that preserves authentic object arrangements and spatial constraints. We design a human-in-the-loop workflow encompassing instance recognition, architectural reconstruction, and object generation and placement, constructing a high-fidelity interactive benchmark derived from 30 real-world residences. Through systematic evaluation across 3D detection, reconstruction, navigation, and mobile manipulation tasks, our results reveal significant limitations of current methods under dense occlusion and confined spaces. This work bridges the gap in interactive, realistic home simulation and advances the deployment of embodied intelligence in authentic domestic environments.

Abstract

Realistic household simulation must capture not only diverse environments but also the lived-in object arrangements and spatial constraints that shape robot motion and interaction. Existing resources often trade off scale, real-world correspondence, and interaction readiness, leaving a gap in faithful, interactive replicas of how real homes are actually arranged. To this end, we introduce LIVIN, a benchmark for spatial and embodied intelligence built on digital twins of 30 diverse lived-in homes. These replicas preserve observed room layouts, furniture configurations, and everyday belongings. To construct them, we design a human-in-the-loop workflow comprising instance recognition, architectural reconstruction, and object generation and placement, with intermediate results reviewed and corrected by humans against the source observations at each stage. We evaluate four tasks in LIVIN: 3D detection, 3D reconstruction, navigation, and loco-manipulation. Our evaluations show that current methods remain challenged by the dense object arrangements, occlusions, limited free space, and constrained interaction regions found in realistic lived-in homes. We hope LIVIN will help advance embodied AI in real-world homes, from spatial understanding to robotic interaction, and ultimately bring embodied intelligence into everyday home environments.

Problem

• Embodied AI
• Digital Twins
• Household Simulation
• Spatial Intelligence
• Benchmarking

Innovation

• Digital Twins
• Embodied Intelligence
• Human-in-the-loop
• Benchmark
• Loco-manipulation