grasping
In plain wordsHow a robot picks up and holds an object securely with a gripper or hand: deciding where to put its fingers or suction cup and how hard to squeeze.
The robotic problem of choosing and executing a configuration of an end effectorend effector, n.: The tool at the end of a robot arm that actually touches the world, such as a gripper, suction cup, welding torch, paint sprayer or robotic hand. on an object — contact locations, approach direction, finger closure and force — so that the object is held stably despite disturbances, often in a way that suits the next step of a task (grasping a mug by its handle in order to pour). Analytical grasp planning uses object geometry and contact mechanics: a grasp achieves force closure if its contacts can resist any external force and torque, and quality metrics score how robustly. Data-driven methods instead predict grasp success from camera or depth images with convolutional networksconvolutional neural network, n.: A kind of neural network built for images and other grid-like data. It slides small pattern detectors across the input, finding edges first, then shapes, then whole objects. trained on large synthetic datasets, such as Dex-Net, or on real trial and error, as in Google's multi-arm grasping experiments and QT-Opt. Suction cups and parallel-jaw grippers dominate industrial picking; tactile sensingtactile sensing, n.: Giving robots a sense of touch, with sensors in fingertips or skin that feel pressure, texture, vibration and slipping, so they can handle objects gently and securely. and closed-loop regrasping help with slippery, transparent or deformable objects. Grasping is one stage of manipulationmanipulation, n.: A robot's use of arms, hands or grippers to change the world around it: picking things up, moving, pushing, opening, assembling or using tools., preceded by perception and followed by transport and placement.
The planner tries the top-ranked suction point first, then falls back to the parallel gripper.
Origin
Analytical grasp theory grew out of 1980s research on multi-fingered hands and contact mechanics, notably Kenneth Salisbury's work on the Stanford/JPL hand and closure analyses by Van-Duc Nguyen and others. Learning-based grasping took off between 2015 and 2018 with deep networks, Dex-Net (Jeffrey Mahler, Ken Goldberg and colleagues, 2017) and large-scale self-supervised data collection.
Usage note
'Grasp' is also a count noun for one candidate hand pose, as in 'grasp detection'. Force closure is a property of where the contacts are, not of how hard the gripper squeezes.
Related
- manipulationmanipulation, n.: A robot's use of arms, hands or grippers to change the world around it: picking things up, moving, pushing, opening, assembling or using tools.
- end effectorend effector, n.: The tool at the end of a robot arm that actually touches the world, such as a gripper, suction cup, welding torch, paint sprayer or robotic hand.
- dexterous manipulationdexterous manipulation, n.: Skilled handling of objects with robot fingers, such as spinning a pen, turning a key or rotating a cube in one hand: the fine finger work people do without thinking.
- tactile sensingtactile sensing, n.: Giving robots a sense of touch, with sensors in fingertips or skin that feel pressure, texture, vibration and slipping, so they can handle objects gently and securely.
- robot learningrobot learning, n.: Teaching robots skills from data and experience, such as watching demonstrations, practicing by trial and error or training in simulation, instead of programming every movement by hand.
- pose estimationpose estimation, n.: Working out the position and orientation of something from images: where a person's joints are and how their body is posed, or exactly how an object or camera is placed and turned in space.
Sources
- Mahler, J. et al. (2017). Dex-Net 2.0: Deep Learning to Plan Robust Grasps with Synthetic Point Clouds and Analytic Grasp Metrics. Robotics: Science and Systems.
- Levine, S. et al. (2018). Learning hand-eye coordination for robotic grasping with deep learning and large-scale data collection. International Journal of Robotics Research 37(4–5).
Mentioned in
end effectorend effector, n.: The tool at the end of a robot arm that actually touches the world, such as a gripper, suction cup, welding torch, paint sprayer or robotic hand., industrial robotindustrial robot, n.: A programmable robot arm used in factories for jobs such as welding, painting, assembling and moving parts, traditionally working inside a safety cage., manipulationmanipulation, n.: A robot's use of arms, hands or grippers to change the world around it: picking things up, moving, pushing, opening, assembling or using tools., dexterous manipulationdexterous manipulation, n.: Skilled handling of objects with robot fingers, such as spinning a pen, turning a key or rotating a cube in one hand: the fine finger work people do without thinking., tactile sensingtactile sensing, n.: Giving robots a sense of touch, with sensors in fingertips or skin that feel pressure, texture, vibration and slipping, so they can handle objects gently and securely.