The DOI matches its Crossref record; 56 of the 56 references it cites were found in a registry or an index; 11 of the 24 sources its claims cite were checked for support.
Ditto in the House: Building Articulation Models of Indoor Scenes through Interactive Perception
Open accessAbstract
Virtualizing the physical world into virtual models has been a critical technique for robot navigation and planning in the real world. To foster manipulation with articulated objects in everyday life, this work explores building articulation models of indoor scenes through a robot's purposeful inter-actions in these scenes. Prior work on articulation reasoning primarily focuses on siloed objects of limited categories. To extend to room-scale environments, the robot has to efficiently and effectively explore a large-scale 3D space, locate articulated objects, and infer their articulations. We introduce an interactive perception approach to this task. Our approach, named Ditto in the House, discovers possible articulated objects through affordance prediction, interacts with these objects to produce articulated motions, and infers the articulation properties from the visual observations before and after each interaction. It tightly couples affordance prediction and articulation inference to improve both tasks. We demonstrate the effectiveness of our approach in both simulation and real-world scenes. Code and additional results are available at https://ut-austin-rpl.github.io/HouseDitto/
Overview
CoChat Summary
TL;DR
Ditto in the House builds scene-level articulation models of indoor scenes by predicting interaction hotspots, physically interacting with articulated objects, and inferring joint parameters from before/after visual observations, with iterative refinement coupling affordance prediction and articulation inference. It reports gains over baselines in simulation and demonstrates a real-world kitchen reconstruction.
Key findings
- Affordance prediction: the self-exploration-trained model without refinement achieved precision 0.66 and coverage 0.60 for prismatic joints, 0.72 for revolute joints >15°, and 0.10 for >30°, versus 3D AffordanceNet’s 0.56 precision, 0.23 prismatic coverage, 0.26 for >15°, and 0.03 for >30°. Iterative refinement raised revolute joints opened >30° from 10% to 55%.
- Articulation inference: the full model achieved Mobile Seg. IoU 0.81, prismatic angle error 0.04, revolute angle error 25.2, and revolute position error 0.04, compared with ScrewNet’s 0.34, 0.28, 46.19, and 0.13, respectively. Ablations: without refinement, revolute angle error was 49.0 and position error 0.06; without contact-region regularity, 31.0 and 0.05.
- Real-world evaluation: the approach was applied without modification to a real kitchen scene using an iPhone 12 Pro LiDAR/camera scan and human interactions, building a scene-level articulation model; no quantitative real-world numbers are reported.
- The abstract reports discovering around 40% more parts than the baseline with higher precision and a 45% absolute segmentation IoU boost compared with the baseline.
Methods
- Approach: an interactive perception pipeline with two stages: affordance prediction using PointNet++ for point-wise binary classification plus non-maximum suppression to extract interaction hotspots, and articulation inference using a Ditto-based network that takes before/after egocentric point clouds and contact-region heatmaps to output joint parameters and part segmentation. An iterative refinement step updates affordance from the estimated articulation model to produce larger articulated motions.
- Data/sample: trained and evaluated in simulation on CubiCasa5K with the iGibson simulator; 2,500 scenes were split into 1,500/500/500 for training, validation, and testing. After cleaning, the test set covers 1,030 objects with 1,736 mobile parts. Affordance supervision was collected by exploration-driven interactions with a simplified virtual robot (a floating ball) and labeled positive/negative based on whether articulated motion was successfully produced; the articulation network was trained with multi-view observations and simulator ground-truth articulation parameters.
- Baselines/metrics: compared against 3D AffordanceNet, ScrewNet, and ablations (without refinement, without regularity). Metrics include precision, coverage for prismatic and revolute joints (>15°, >30°), Mobile Seg. IoU, joint angle error, and joint position error. Real-world evaluation used an iPhone 12 Pro LiDAR/camera and human hands.
Limitations
- Stated: experiments use simulated robot grippers and human interactions to simplify exploration and manipulation. Exploration limitations: the robot is teleported between locations, abstracting away navigation and motion planning, and perfect odometry and depth estimation are assumed when reconstructing the static model. Interaction limitations: all interactions are performed by creating pseudo links between the robot and objects or by humans, and joint constraints or robot self-collisions are not considered.
- Evident from the text: the real-world evaluation does not use a physical robot; it uses an iPhone 12 Pro scan and human hands, so real robot navigation, manipulation, and collision constraints remain untested, and no quantitative real-world results are reported.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
4 Valid · 2 Not valid · 8 Not checked · 38 Own results and interpretations
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The extracted interaction hotspot can be close to the axis.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our approach also gives a 45% absolute segmentation IoU boost compared with the baseline.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper states — it cites no source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“In our experiments, only 10% of the revolute joints are opened up to > 30◦ during the initial interaction”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Empirically, we find that articulation estimation of a fully opened revolute joint, e.g., > 30◦, is more accurate in terms of angle error than one with an ajar joint.”
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“The data distribution is imbalanced due to the large proportion of negative data.”
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
13 more claims
Create a free account to see all 52 claims and every source verdict.
- Probably not a substantial claim. 20
- It may state two separate claims as one. 16
- The quote only partly states this claim. 3
- Not yet classified. 1
Citations
Cites
Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation
Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2014) · 2014 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2014.81 · Cited by 32,684 · cited 2×
“Since dense affordance prediction may introduce too many potential locations to probe, we apply non-maximum suppression (NMS) [11] to extract peaks from each region as the interaction hotspots.”
MethodKinectFusion: Real-time dense surface mapping and tracking
Newcombe, R. A., Fitzgibbon, A., Izadi, S., Hilliges, O., Molyneaux, D., Kim, D., Davison, A. J., Kohi, P., Shotton, J., & Hodges, S. (2011) · 2011 10th IEEE International Symposium on Mixed and Augmented Reality · doi:10.1109/ismar.2011.6092378 · Cited by 4,008 · cited 1×
“Recent advances in structure sensors and SLAM algorithms [9, 29, 30] have offered ways to construct static replicas of realworld scenes at unprecedented fidelity.”
BackgroundPointNet++: Deep Hierarchical Feature Learning on Point Sets in a Metric Space
Qi, C. R., Yi, L., Su, H., & Guibas, L. J. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1706.02413 · Cited by 2,114
DTAM: Dense tracking and mapping in real-time
Newcombe, R. A., Lovegrove, S. J., & Davison, A. J. (2011) · 2011 International Conference on Computer Vision · doi:10.1109/iccv.2011.6126513 · Cited by 2,078
StereoScan: Dense 3d reconstruction in real-time
Geiger, A., Ziegler, J., & Stiller, C. (2011) · 2011 IEEE Intelligent Vehicles Symposium (IV) · doi:10.1109/ivs.2011.5940405 · Cited by 1,089 · cited 1×
“Recent advances in structure sensors and SLAM algorithms [9, 29, 30] have offered ways to construct static replicas of realworld scenes at unprecedented fidelity.”
BackgroundDynamicFusion: Reconstruction and tracking of non-rigid scenes in real-time
Newcombe, R. A., Fox, D., & Seitz, S. M. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298631 · Cited by 951 · cited 1×
“Recent advances in structure sensors and SLAM algorithms [9, 29, 30] have offered ways to construct static replicas of realworld scenes at unprecedented fidelity.”
BackgroundLearning Synergies Between Pushing and Grasping with Self-Supervised Deep Reinforcement Learning
Zeng, A., Song, S., Welker, S., Lee, J., Rodriguez, A., & Funkhouser, T. (2018) · 2018 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros.2018.8593986 · Cited by 606
Neural Body Fitting: Unifying Deep Learning and Model Based Human Pose and Shape Estimation
Omran, M., Lassner, C., Pons-Moll, G., Gehler, P., & Schiele, B. (2018) · International Conference on 3D Vision · doi:10.48550/arxiv.1808.05942 · Cited by 533
Neural Body Fitting: Unifying Deep Learning and Model Based Human Pose and Shape Estimation
Omran, M., Lassner, C., Pons-Moll, G., Gehler, P., & Schiele, B. (2018) · 2018 International Conference on 3D Vision (3DV) · doi:10.1109/3dv.2018.00062 · Cited by 498
The Ecological Approach to the Visual Perception of Pictures
Gibson, J. J. (1978) · Leonardo · doi:10.2307/1574154 · Cited by 456
Cited by
SceneFun3D: Fine-Grained Functionality and Affordance Understanding in 3D Scenes
Delitzas, A., Takmaz, A., Tombari, F., Sumner, R. W., Pollefeys, M., & Engelmann, F. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.01377 · Cited by 50
“[26, 31, 36, 40, 53] explores the estimation of 3D motion and mobility of in-teractable elements.”
BackgroundPARIS: Part-level Reconstruction and Motion Analysis for Articulated Objects
Liu, J., Amiri, A. M., & Savva, M. (2023) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51070.2023.00039 · Cited by 55
“Acquiring digital replicas of articulated objects [20, 16] also enables simulating object articulation in applications involving robotic agents and embodied AI [37, 41].”
BackgroundAffordanceLLM: Grounding Affordance from Vision Language Models
Qian, S., Chen, W., Bai, M., Zhou, X. H., Tu, Z., & Li, L. E. (2024) · IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW) · doi:10.1109/cvprw63382.2024.00754 · Cited by 140
“…usually available for an arbitrary image, fortunately, researchers have built a series of robust approaches to estimate the 3D of an image, ranging from surface normal estimation [15, 61], depth estimation [10, 33, 55, 68, 69], to 3D reconstruction [18, 35, 47] and few-image NeRF [5, 37, 52, 70].”
BackgroundRecent Trends in 3D Reconstruction of General Non‐Rigid Scenes
Yunus, R., Lenssen, J. E., Niemeyer, M., Liao, Y., Rupprecht, C., Theobalt, C., Pons‐Moll, G., Huang, J., Golyanik, V., & Ilg, E. (2024) · Computer Graphics Forum · doi:10.1111/cgf.15062 · Cited by 38
Neural Implicit Representation for Building Digital Twins of Unknown Articulated Objects
Weng, Y., Wen, B., Tremblay, J., Blukis, V., Fox, D., Guibas, L. J., & Birchfield, S. T. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.00303 · Cited by 35
Metadata
- Authors
- Hsu, C.-C., Jiang, Z., & Zhu, Y.
- Affiliations
- The University of Texas at Austin (US)
- Venue
- 2023 IEEE International Conference on Robotics and Automation (ICRA)
- Publisher
- IEEE
- Published
- 2023-05-29
- Pages
- 3933-3939
- ISSN
- 1050-4729 (ISSN-L)
- OpenAlex
- W4383109308
- Cited by
- 21
- References
- 54
- Citations by year
- 2026: 2 · 2025: 7 · 2024: 6 · 2023: 6
- FWCI
- 8.99
- Type
- conference paper
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- Proceedings - IEEE International Conference on Robotics and Automation/Proceedings (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Funders
- NSF (CNS-1955523,FRR-2145283)
- Topics
- Physical Sciences › Engineering › Control and Systems Engineering › Robot Manipulation and Learning
- Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Robotics and Sensor-Based Localization
- Keywords
- indoor scene understanding, interactive perception, affordance prediction, robotic manipulation, robot exploration, 3D object modeling
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-01, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-01 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Ditto in the House: Building Articulation Models of Indoor Scenes through Interactive Perception. https://app.cochat.ai/papers/10.1109/icra48891.2023.10161431
- The paper
- Hsu, C.-C., Jiang, Z., & Zhu, Y. (2023). Ditto in the House: Building Articulation Models of Indoor Scenes through Interactive Perception. 2023 IEEE International Conference on Robotics and Automation (ICRA). https://doi.org/10.1109/icra48891.2023.10161431
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.