ShadowSense: Detecting Human Touch in a Social Robot Using Shadow Image Classification
summary
A camera sits inside a translucent skin and captures touch-generated silhouettes that a shadow image classification pipeline converts into gesture labels and positions, implemented on an inflatable prototype with a fisheye lens. Gesture classifier uses Densely Connected Convolutional Networks trained on 6000 images of six gestures (hug, palm, point, punch, two‑hand, none) across daylight/dusk/night and achieves about 87–96% accuracy; a color‑segmentation tracker localizes touch and supports hover, interactive projection, and privacy‑preserving sensing.
AI-written summary · about
related talks
12mEclipseTouch: Touch Segmentation on Ad Hoc Surfaces using Worn Infrared Shadow Casting
18mCyclopsRing: Enabling Whole-Hand and Context-Aware Interactions Through a Fisheye Ring
17mEchoFlex: Hand Gesture Recognition using Ultrasound Imaging
05mReady, Steady, Touch! — Sensing Physical Contact with a Finger-Mounted IMU
18mSelf-Powered Gesture Recognition with Ambient Light
19mSymmetriSense: Enabling Near-Surface Interactivity on Glossy Surfaces using a Single Commodity ...
13mEnhancing Object Detection & Avoidance Systems for Robots in Challenging Environments Using AI
10mDetecting 3D Hand Pointing Direction from RGB-D Data in Wide-Ranging HRI Scenarios