Mark Weiser Best Paper Award

Wandatch: Infrastructure-Free Point-to-Command with Smartwatches and Speakers

Lin Chen, Yandao Huang, Minghui Qiu, Shuxin Zhong, Jun Chen, Kaishun Wu

PerCom, 2026

Smart home scenarios controlled by Wandatch — Wandatch turns a smartwatch and speaker into a wrist-ray interface for appliance control.

A visual tour of how Wandatch enables point-and-command smart-home interaction without cameras, tags, anchors, or room-specific calibration.

97.6%selection accuracy in sparse real-appliance deployments

85.5%selection accuracy with appliances spaced only 40 cm apart

12.6-50.1%shorter interaction time than app and voice baselines

0extra infrastructure devices attached to appliances

Wandatch system overview from activation to gesture control

Workflow

From wrist motion to appliance command

The interaction is organized as a three-step loop: activate the watch with a micro gesture, confirm the target through acoustic localization and IMU orientation, then control it with single- or multi-stroke gestures.

Gesture Layer

Gestures are treated as an interface

The source sketches make the command vocabulary readable immediately: up, down, left, right, and rotate become the atomic controls for appliances.

Micro gestures wake and select without a rigid gesture-pause routine.
Single-stroke gestures handle directional commands such as up, down, left, right, and rotate.
Multi-stroke recognition keeps richer air-writing commands practical.

Gyroscope readings for five single-stroke gestures

Gesture Signals

Motion traces reveal why the gestures are separable

The gyroscope traces show distinct roll, pitch, and yaw patterns for directional commands, making the recognition logic more inspectable than a paragraph-only explanation.

Deployment

Sparse deployment tests room-scale pointing

The office layout shows where participants stood, where appliances were placed, and why a wrist-ray interface needs reliable spatial discrimination.

Deployment

Dense deployment tests close-range ambiguity

The dense setup stresses selection accuracy when appliances are arranged close together, including spacing down to 40 cm.

Appliance selection accuracy for different users

Evaluation

Selection accuracy remains high across users

The per-user accuracy plot makes the core result easy to scan before opening the PDF for full experimental details.

Interaction time comparison across Wandatch, mobile app, and voice control

Efficiency

Wandatch reduces interaction time

The time comparison visually explains the 12.6-50.1% improvement over smartphone app and voice-control baselines.

Smartwatch power consumption comparison across operating states

Efficiency

The smartwatch cost is visible too

Power consumption is shown separately so the page does not hide the wearable-side tradeoff behind a single performance number.

CDF of 3D positioning errors in an open environment

Localization

Localization performance is summarized as error CDFs

The positioning CDFs expose how much error comes from each axis, turning the acoustic localization result into an at-a-glance visual.

CDF of 3D positioning errors in an office room

Localization

Office-room multipath is shown separately

The second CDF makes the environment shift explicit, helping readers distinguish open-space behavior from a more realistic office setting.

User Study

Participants rated Wandatch higher on most usability dimensions

The questionnaire summary highlights where the interaction felt better than smartphone apps and voice control, especially around speed, controllability, enjoyment, and preference.

Citation

@inproceedings{chen2026wandatch,
  title = {Wandatch: Infrastructure-Free Point-to-Command with Smartwatches and Speakers},
  author = {Chen, Lin and Huang, Yandao and Qiu, Minghui and Zhong, Shuxin and Chen, Jun and Wu, Kaishun},
  booktitle = {2026 IEEE International Conference on Pervasive Computing and Communications (PerCom)},
  year = {2026}
}

Lin CHEN (Leen)