Unisoma: A Unified Transformer-based Solver for Multi-Solid Systems

2025-06-06

AI Summary

Modeling physical interactions in multi-solid systems remains challenging, and existing deep learning approaches suffer from sharp accuracy degradation as the number of solids increases. Method: This paper introduces the first explicit, structured Transformer framework tailored for variable-scale multi-rigid-body/multi-solid systems. It innovatively incorporates a contact-aware module and an adaptive interaction allocation mechanism, explicitly decoupling inter-solid couplings via ternary relational modeling—thereby avoiding information entanglement inherent in implicit modeling. The framework performs end-to-end prediction of solid deformations while ensuring physical interpretability and cross-scale generalizability. Contribution/Results: Evaluated on seven standard benchmarks and two complex multi-solid tasks, the method achieves state-of-the-art performance, significantly improving both prediction accuracy and generalization across varying system scales.

Technology Category

• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Machine Learning: Multimodal Learning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

Multi-solid systems are foundational to a wide range of real-world applications, yet modeling their complex interactions remains challenging. Existing deep learning methods predominantly rely on implicit modeling, where the factors influencing solid deformation are not explicitly represented but are instead indirectly learned. However, as the number of solids increases, these methods struggle to accurately capture intricate physical interactions. In this paper, we introduce a novel explicit modeling paradigm that incorporates factors influencing solid deformation through structured modules. Specifically, we present Unisoma, a unified and flexible Transformer-based model capable of handling variable numbers of solids. Unisoma directly captures physical interactions using contact modules and adaptive interaction allocation mechanism, and learns the deformation through a triplet relationship. Compared to implicit modeling techniques, explicit modeling is more well-suited for multi-solid systems with diverse coupling patterns, as it enables detailed treatment of each solid while preventing information blending and confusion. Experimentally, Unisoma achieves consistent state-of-the-art performance across seven well-established datasets and two complex multi-solid tasks. Code is avaiable at href{this link}{https://github.com/therontau0054/Unisoma}.

Problem

• Modeling complex interactions in multi-solid systems
• Overcoming limitations of implicit deformation modeling
• Handling variable solid counts with explicit physical interactions

Innovation

• Unified Transformer-based explicit modeling
• Contact modules for physical interactions
• Adaptive interaction allocation mechanism