HoldLight lets climbers hear the route before the next move

“Right hand, two o’clock, far.” HoldLight’s brief cue follows a larger task: helping a blind or low-vision climber understand the route before leaving the ground. Its preprint was posted on 9 October.[1]

The work is listed in the ASSETS conference’s 26 October session on mobility and exercise.[3]

The prototype scans holds, constructs a route with a rule-based planner, and offers an overview before optional detail. During climbing, body tracking and wall alignment support selective spoken cues. The phone stays on the ground.[1]

Choosing what reaches the climber

An earlier project, Climb-o-Vision, makes the sensory choice unusually clear. In their 2022 paper, University of Bath researchers Mike Richardson, Karin Petrini and Michael Proulx used a trained object detector to locate holds and represent their positions through electrodes on the tongue. Keeping hearing available was part of the design.[2]

The software narrowed the camera’s view to a horizontal strip and supplied lateral information. The climber would move their head to explore height. This avoided demanding a detailed vertical image from a tongue display whose sensitivity varies across its surface. Recognition filtered the scene to the task; physical exploration supplied another dimension.[2]

The researchers tested detection after the gym’s routes were reset and found troublesome colour confusions. Measuring blind climbers’ performance was still future work. Their contribution was a prototype for selecting what information should reach a different sense.[2]

Twelve climbers compared HoldLight with a guide restricted to requests or major pauses during climbs. Median pre-climb route-understanding scores were 2 with the guide and 3.5 with HoldLight, on a 0–4 rubric. The comparison covered the whole workflow, including preview, with human safety support throughout.[1]

The study found no detectable change in task time. In three of 36 trials, a mistaken route anchor produced a misleading cue and human help was needed. Its encouraging result concerns understanding the route, within that supported setting.[1]

Report an error or suggest a correction