Contact-Free Grasp Stability Prediction with In-Hand Time-of-Flight Sensors
New method predicts grasp stability before contact, cycling at 15 Hz without touching the object.
Deep Dive
Researchers Kyle DuFrene and Cindy Grimm propose a contact-free grasp stability predictor using multi-zone time-of-flight sensors mounted in a gripper's distal links. Unlike tactile methods, it doesn't require grasping to predict stability, cycling at 15 Hz. Trained on 2,500 real-world grasps across 15 objects, it achieved 85.5% validation and 86.0% test accuracy on unseen objects.
Key Points
- Uses multi-zone time-of-flight sensors in gripper distal links for contact-free prediction.
- Classifier operates at 15 Hz, trained on 2,500 grasps across 15 objects.
- Achieved 86% accuracy on unseen test objects without any physical contact.
Why It Matters
Contact-free grasp prediction enables faster, safer robotic manipulation by detecting failures before grasping.