NoContactNoWorries: Estimating Contact through Vision and Proprioception for In-Hand Dexterous Manipulation

A team of robotics researchers has introduced a framework that allows a robot hand to infer physical contact without dedicated tactile sensors, using only camera data and internal joint-position readings. The system, called NoContactNoWorries, fuses RGB-D vision with proprioception to generate a pseudo-tactile signal for in-hand manipulation tasks [1]. The work, posted to the arXiv preprint repository on June 23, 2026, addresses a persistent hardware bottleneck in dexterous manipulation. Tactile sensors, while valuable, present practical challenges in cost, fragility, and integration [2]. The researchers instead draw inspiration from human embodied perception, where visual information combines with an innate sense of body pose and movement to infer contact [1]. The NoContactNoWorries framework uses a transformer-based multimodal architecture to fuse RGB-D camera streams with the robot's proprioceptive data, outputting binary contact states [2]. This signal acts as a stand-in for tactile hardware, specifically for binary contact estimation, and is designed to feed downstream reinforcement learning agents that perform in-hand object reorientation [1]. The team trained a single contact prediction model on multiple objects and demonstrated that the inferred contact signal generalizes to novel objects, with experiments conducted in both simulation and on a real-world robot platform [2]. The preprint appears on arXiv, an open-access repository that has hosted over two million articles since its founding in 1991 and currently receives roughly 24,000 submissions per month [6]. The paper is accompanied by a project page and is accessible through the repository's standard abstract page, which also links to community-developed tools such as the Bibliographic Explorer and Connected Papers [5][4]. These tools are part of the arXivLabs framework, a formalized collaboration space that allows third-party developers to build experimental features on top of arXiv's article records while adhering to the repository's values of openness, community, excellence, and user data privacy [4].