Proximity3D: Shape from Capacitive Proximity on Sensing Manifold

Hao Chen*,1,2,Chenming Wu*,Chun Ping Lam1,Xiangjia Chen1,Guoxin Fang3,1,Charlie C. L. Wang4,Yeung Yam3,1,Juncong Lin†,2,Chengkai Dai†,1
*Equal contribution†Corresponding authors
1Centre for Perceptual and Interactive Intelligence2Xiamen University3The Chinese University of Hong Kong4The University of Manchester
▤Paper</>Code
Overview of the Proximity3D pipeline
Proximity3D reconstructs object geometry from pre-contact capacitive scans captured by a curved woven sensing manifold. Multi-view proximity fields are fused to recover a complete 3D mesh for downstream robotic applications.

Abstract

Most shape reconstruction methods assume measurements defined over planar sensing domains, such as RGB images or depth maps. In this paper, we use a curved capacitive textile as a shape sensor, treating its surface as a non-planar sensing manifold. Each scan is represented as a capacitive proximity field on this manifold, induced by the interaction between the curved electrode layout and nearby object geometry.

We introduce a multi-view feedforward reconstruction model that aggregates these fields across known sensor views and recovers the observed object shape. Simulated and physical experiments demonstrate robust reconstruction from capacitive proximity signals acquired on curved sensing surfaces, pointing toward a new route to robotic near-field geometric awareness via embodied sensing.

Capacitive Proximity on a Sensing Manifold

A woven capacitive textile conforms to a curved surface. Each channel has a known site, local electrode frame, and adjacency relation, providing the geometric structure used by the reconstruction model.

Curved woven capacitive sensing manifold and acquisition setup
The sensing manifold is moved around the target without contact. Each view yields a capacitive proximity field defined on the curved electrode layout.

Method

Proximity3D method architecture
Manifold Sensing Attention aggregates local channel responses with the local electrode layout and tangent-frame geometry. A global module then fuses evidence across known views into a 3D latent representation for shape decoding.

Simulated Reconstruction Results

Ground Truth

Reconstruction

Real-World Results

The woven sensing manifold acquires multi-view proximity signals around each physical target, which Proximity3D uses to reconstruct the object mesh. The figure below is a schematic illustration of the real-world sensing process.

Real-world capacitive sensing sequences and physical target objects

BibTeX

@misc{chen2026proximity3dshapecapacitiveproximity,
  title = {Proximity3D: Shape from Capacitive Proximity on Sensing Manifold},
  author = {Hao Chen and Chenming Wu and Chun Ping Lam and Xiangjia Chen and
            Guoxin Fang and Charlie C. L. Wang and Yeung Yam and Juncong Lin and
            Chengkai Dai},
  year = {2026},
  eprint = {2608.30344},
  archivePrefix = {arXiv},
  primaryClass = {cs.CV},
  url = {https://arxiv.org/abs/2608.30344}
}