Ankush Dhawan

Aligns human demonstrations from a capacitive tactile glove with robot data across vision, touch, and kinematics to co-train a visuotactile diffusion policy. Adding aligned human demos lifts Lego assembly success from 10% to 80% and enables strawberry plucking across all fruit sizes.

A triangular roller tip mount that carries sensors and tools on soft vine robots growing through tip eversion, with a benchmarking testbed that characterizes how mounting approaches affect growth speed and internal resistance.

Transformer-based cloth layer detection using DenseTact 2.0 optical tactile sensors. Achieves 98.21% accuracy in classifying grasped cloth layers through anthropomorphic rubbing motion.

Experiments with the SPRoUT vine robot in engineered collapse structures for urban search and rescue scenarios. Tests viability, control, and sensorization of soft growing robots.

Soft optical tactile fingertip enabling tap, rolling, and fingernail grasping strategies. Demonstrates >90% success rate across small and thin objects including 1 mm basil seeds and thin paperclips.

A core-double shell composite consisting of a sand base substrate with sequential layers of chitosan and graphene oxide for Pb(II) adsorption from aqueous solutions. The material achieved lead concentrations below the EPA limit (<15 ppb), with adsorption characteristics following the Freundlich isotherm model.

A test kit for detecting trace levels of arsenic in water samples using a stable bromide compound, dye, and acid solution. Capable of detecting concentrations up to 0.03 parts per million with accuracy and repeatability, meeting US EPA and WHO standards for potable water testing.

A microbial fuel cell light assembly that generates electricity through microbial activity, providing sustainable lighting for a range of applications by harnessing biological processes to create clean energy.