Demonstrating Eyes-Free Object Retrieval via Fingertip Deflection Guidance Using the NURing
Tomasz P. Trzpit, Gregory Reardon, Elizabeth Gerber, Pedro Lopes, Michael Peshkin, J. Edward Colgate · 2026 · Extended Abstracts of the 2026 CHI Conference on Human Factors in Computing Systems (CHI EA ’26) · doi:10.1145/3772363.3799146
Summary
This CHI 2026 Extended Abstract is an interactive-demo paper introducing a vision-enabled version of NURing, a tendon-driven wearable ring that guides reach by gently deflecting the wearer’s fingertip. The authors frame the work around a recurring report from blind and low-vision (BVI) users: that retrieving small objects within arm’s reach — a dropped pill, a utensil — is one of the most time-consuming and frustrating tasks of everyday life, and one that can be safety-critical in homes shared with children or pets. They argue most existing eyes-free guidance systems either occupy hearing (spatialised audio, spoken commands) or deliver vibrotactile alerts that have to be cognitively decoded into directions, slowing movement. Inspired by the white cane and the everyday gesture of a child leading an adult by the finger, NURing instead applies continuous kinesthetic deflection to the index finger near the proximal interphalangeal joint, using three independently actuated tendons routed from a forearm brace. The fingertip pad stays free for natural touch, and deflection forces are deliberately small — perceptible but overridable, preserving user agency. The new contribution over prior NURing work is a wide-angle 120° global-shutter camera mounted on the actuation ring itself, which uses OpenCV/ArUco to detect tagged targets at 60 Hz, estimate their 3D pose relative to the camera, and feed the error vector into a proportional controller that drives closed-loop deflection during free reach rather than on a constrained pivot testbed. The demo asks attendees to retrieve five 25-mm cubes from a vertical board with their eyes closed, deposit them in a tray, and run the loop until the board is clear.
Key findings
As a demo paper rather than a study paper, this contribution is engineering and experiential rather than statistical. The headline technical result is that fingertip-deflection guidance can be lifted out of a sensorised testbed and into free reach using only a ring-mounted camera and ArUco tags, with a 270 g total wearable weight and a 1600×1200 px / 60 Hz vision pipeline. The authors describe two new haptic primitives beyond directional deflection: a "stop" cue that pulls all three tendons at once to render a virtual wall when tracking is lost, and a sharper pull-and-release pulse that causes users to retract and reacquire the target. They position the camera-on-ring approach as a stepping stone toward markerless guidance via object recognition, scene understanding, or SLAM, with ArUco tags chosen for repeatability rather than as an end-state. The argument running through the paper is methodological: the white cane has outlasted decades of sensor-rich alternatives because it gives continuous, embodied feedback rather than symbolic cues, and assistive guidance should aim for the same property. The demo itself takes under five minutes per attendee, with the retrieval task itself under a minute, and collects open feedback rather than standardised measures.
Relevance
For accessibility practitioners this is a useful counter-example to the dominant "encode information into tactile patterns" pattern in BVI guidance research. The design principle — bias the body, do not narrate directions — is portable: it could inform robot-assisted manipulation, surgical training, AR pointing tasks, and motor rehabilitation. The form factor is also notable: by deflecting at the PIP joint the design leaves the fingertip free for braille, texture, and grasp, which is a hard constraint that many haptic gloves miss. Limitations are significant. There is no user study with BVI participants — the paper is explicit that the demo is an experiential preview, not an evaluation, and the only motivating BVI evidence is cited from prior accessibility literature. ArUco tags are an instrumentation crutch that real homes will not have, and the authors flag this. The device requires a forearm brace, hook-and-loop straps, and facilitator setup; it is not yet a self-don wearable. And the 270 g mass on the dominant forearm may be tiring over long sessions. Anyone planning to evaluate this with BVI users should also engage on training time, interaction with white-cane and guide-dog use, and what happens when multiple targets are in the camera frame.
Tags: haptic feedback · kinesthetic feedback · wearable technology · blindness · visual impairment · assistive technology · eyes-free interaction · guidance