🙏 This work highlights the advantages of soft robots in assistive care. Thanks to all collaborators: @ndennler.bsky.social Eliot Xing Maja Matarić Stefanos Nikolaidis @jeff-ichnowski.bsky.social Jean Oh @HRI_Conference 🧵7/7

🚀 We've already extended MOE to various dexterity projects including dynamic pen spinning (arxiv.org/abs/2411.12734) and learning in-hand manipulation from demonstration (arxiv.org/abs/2503.01078). 🧵6/7

👥 Our study with 12 participants significantly preferred MOE with force estimation-based feedback across all tasks! Feedback included: "It felt really similar to human fingers" and felt "like a head massage.”🧵5/7

🔍 The force estimation module combines visual deformation data 👁️ with tendon tensions 📈to precisely track applied forces, reducing sensing errors by up to 60% compared to tension-only approaches. 🧵4/7

Not surprisingly, MOE passively applies 74% less force than rigid grippers while grasping comparable amounts of hair. This implies MOE is capable of more comfortable care without sacrificing effectiveness.🧵3/8

💡 Rigid robots feel "rough" on hair and struggle with safety when hair obscures the scalp. 🐙🤖 We introduce MOE: a tendon-driven soft robot hand. Our system's inherent compliance provides both comfort and safer interaction. 🧵2/7

Thanks! It's a dexterous soft hand that we're calling MOE. It has silicone rubber fingers with 4 tendons each that are pulled with motors to make them bend. We're preparing to open source it soon!