A proposed AAAI-27 workshop

Geometric & Spatial Reasoning in Vision-Language Models

FORMATFull-day workshopLOCATIONMontréal, CanadaWORKSHOP DATETo be announced

Overview

Vision-language models struggle to maintain geometric consistency across fragments, viewpoints, and reasoning steps. This workshop examines these limitations and methods for improving spatial reasoning as task complexity increases.

The program brings together researchers in multimodal learning, embodied AI, and constraint reasoning through invited talks, contributed papers, and posters.

01

Spatial perception

Recognize relations, depth, and orientation already present in an image.

02

Spatial construction

Construct and maintain globally consistent structures under geometric constraints.

Topics of interest

01Spatial perception & scene understanding+

Spatial relationships between objects, depth, size, orientation, and scene understanding in 2D and 3D.

02Geometric reasoning & spatial problem solving+

Geometric relationships, spatial puzzles, object assembly, and reasoning about how objects fit and move in space.

03Model architectures for spatial reasoning+

Model designs for representing spatial information, combining vision and language, and reasoning across multiple views.

04Neuro-symbolic & tool-assisted reasoning+

Combining vision-language models with symbolic reasoning, search, planning, and external tools to solve spatial tasks.

05Learning spatial representations+

Pretraining, fine-tuning, reinforcement learning, and data generation for improving geometric and spatial reasoning.

06Benchmarks & evaluation+

Datasets, benchmarks, and metrics for evaluating spatial understanding, reasoning accuracy, robustness, and generalization.

07Embodied AI & applications+

Spatial reasoning for robotics, navigation, manipulation, and interaction with real or simulated environments.

Invited speakers

Mohit Bansal

Mohit Bansal

University of North Carolina at Chapel Hill

Joydeep Biswas

Joydeep Biswas

University of Texas at Austin

Scott Sanner

Scott Sanner

University of Toronto

Ashok Goel

Ashok Goel

Georgia Institute of Technology

Ferdinando Fioretto

Ferdinando Fioretto

University of Virginia

Anthony Cohn

Anthony Cohn

University of Leeds

Call for papers

We invite research and position papers on the topics above.

Long and short/position papers must be anonymized for double-blind review.

This workshop is non-archival.

Long papers

Up to 7 pages of main content, with no page limit for references and appendices.

7-page main text

Short & position papers

Up to 4 pages of main content and up to 2 pages of references.

4-page main text

Every accepted paper will be presented as a poster. Selected papers will also receive an oral presentation slot.

Program

Tentative schedule
WELCOME

Opening remarks & problem framing

CONTRIBUTED RESEARCH

Oral presentations I

Four talks · 15 minutes each

POSTER SESSION

Coffee break & poster session I

AWARD & PHOTO

Best paper award & group photo

Organizing committee

Shawn Li

Shawn Li

University of Southern California

Yue Zhao

Yue Zhao

University of Southern California

Leman Akoglu

Leman Akoglu

Carnegie Mellon University

Xiyang Hu

Xiyang Hu

Arizona State University

Emilio Ferrara

Emilio Ferrara

University of Southern California

SUPPORTING TEAM
Tianyu Shi

Tianyu Shi

McGill University

Ruolin Li

Ruolin Li

University of Southern California