UniForce: A Unified Latent Force Model for Robot Manipulation with Diverse Tactile Sensors

2026-02-01

AI Summary

This work addresses the challenge of achieving universal force perception across heterogeneous tactile sensors, which differ significantly in sensing mechanisms and physical morphology. To overcome this limitation, the authors propose UniForce, a framework that jointly models inverse (image-to-force) and forward (force-to-image) dynamics and constructs a shared latent force space across sensors under static force equilibrium constraints. This approach enables cross-device alignment without requiring external force/torque sensors. Notably, UniForce achieves zero-shot force perception transfer among multimodal tactile sensors for the first time, supporting plug-and-play deployment in downstream tasks and facilitating vision–touch–language–action coordinated manipulation. Experiments demonstrate substantial improvements in force estimation accuracy across diverse platforms—including GelSight, TacTip, and uSkin—and successful application in a robot wiping task requiring cross-sensor coordination.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Computer Vision: Multi-modal Vision
• Humans and AI: Interaction Techniques and Devices

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Force sensing is essential for dexterous robot manipulation, but scaling force-aware policy learning is hindered by the heterogeneity of tactile sensors. Differences in sensing principles (e.g., optical vs. magnetic), form factors, and materials typically require sensor-specific data collection, calibration, and model training, thereby limiting generalisability. We propose UniForce, a novel unified tactile representation learning framework that learns a shared latent force space across diverse tactile sensors. UniForce reduces cross-sensor domain shift by jointly modeling inverse dynamics (image-to-force) and forward dynamics (force-to-image), constrained by force equilibrium and image reconstruction losses to produce force-grounded representations. To avoid reliance on expensive external force/torque (F/T) sensors, we exploit static equilibrium and collect force-paired data via direct sensor--object--sensor interactions, enabling cross-sensor alignment with contact force. The resulting universal tactile encoder can be plugged into downstream force-aware robot manipulation tasks with zero-shot transfer, without retraining or finetuning. Extensive experiments on heterogeneous tactile sensors including GelSight, TacTip, and uSkin, demonstrate consistent improvements in force estimation over prior methods, and enable effective cross-sensor coordination in Vision-Tactile-Language-Action (VTLA) models for a robotic wiping task. Code and datasets will be released.

Problem

• tactile sensors
• force sensing
• domain shift
• robot manipulation
• sensor heterogeneity

Innovation

• unified tactile representation
• latent force space
• cross-sensor alignment
• zero-shot transfer
• force-grounded dynamics modeling