Paper
CHORD: Contact-Guided Dexterous Manipulation
NVIDIA Isaac's CHORD teaches robots dexterous, two-handed manipulation by learning from human demonstrations.
Paper
T-Rex: Tactile-Reactive Dexterous Manipulation
UC Berkeley, NVIDIA, and Stanford give robots a sense of touch, boosting success on delicate two-handed manipulation tasks.
Paper
DUET: Dual-Robot Learning via Efficient Teaching
USC’s DUET teaches two Dexmate Vega robots to collaborate by watching pairs of humans work together.
Paper
PGDG: Robust Bimanual Policies from One Demonstration
Dexmate and Carnegie Mellon turn one demonstration into a full training dataset, teaching robots to recover from mistakes.
Paper
ResFiT: Residual Off-Policy RL for Fine-Tuning Robot Policies
Amazon FAR fine-tunes behavior cloning with lightweight off-policy RL on Dexmate's Vega for real-world dexterous manipulation.