Projects

Research on haptics, robotics, and human-computer interaction, newest first.

Turn a 30-second phone video into a touchable twin, then fix what feels wrong by typing what your hand tells you.

Estimating contact from tactile video and force, and rendering it to the operator's hand during teleoperation.

Putting measured materials, not random friction values, into simulation, so robot hands learn how real surfaces grip, slide, and slip.

Type “gritty but cushioned” and feel it: one prompt generates sliding vibration, tapping, and a matching visual.

Authoring shape-changing behaviors on a 24 × 24 pin display through natural-language commands.

An early look at SHAPE-IT: GPT-4 turns natural-language instructions into dynamic shapes on a pin-based display.

A weakly supervised, part-based detector for combined objects in remote sensing images that needs less labeling.

A cable-driven robot that reproduces human graffiti painting motions, from motion capture to spray paint.

Transfers lighting from a reference image to a target image with depth guidance, for the NTIRE 2021 relighting challenge.

Adapts the number of stroke control points to each character's complexity, so generated calligraphy follows the original more closely.