Grasp selection
AssigneeHonda Motor Co., Ltd.
InventorsThomas Power, Rana Soltani Zarrin
Life of this patent
Estimating the term…
Abstract
Systems and techniques for grasp selection may include receiving one or more candidate object trajectories and a current grasp of a robotic hand on an object, sampling random candidate grasps for the one or more candidate object trajectories based on the current grasp, generating one or more grasps to be optimized for each of the one or more candidate object trajectories based on the sampled candidate grasps, and optimizing one or more of the grasps to be optimized for each of the one or more candidate object trajectories based on a cost function.
Claims20
A system for grasp selection, comprising:
- a memory storing one or more instructions;
- and
- a processor executing one or more of the instructions stored on the memory to perform:
- receiving one or more candidate object trajectories and a current grasp of a robotic hand on an object;
- sampling random candidate grasps for the one or more candidate object trajectories based on the current grasp of the robotic hand;
- generating one or more grasps to be optimized for each of the one or more candidate object trajectories based on the sampled candidate grasps;
- and
- optimizing one or more of the grasps to be optimized for each of the one or more candidate object trajectories based on a neural network cost function of a combination of a grasp quality cost and a task-specific cost by using gradient descent to minimize the neural network cost function.
A system for grasp selection, comprising:
- a memory storing one or more instructions;
- and
- a processor executing one or more of the instructions stored on the memory to perform:
- receiving one or more candidate object trajectories and a current grasp of a robotic hand on an object, wherein each one of the one or more candidate object trajectories includes two or more waypoints;
- sampling random candidate grasps for the one or more candidate object trajectories based on the current grasp of the robotic hand;
- generating one or more grasps to be optimized for each waypoint of each of the candidate object trajectories based on the sampled candidate grasps;
- and
- optimizing one or more of the grasps to be optimized for each of the one or more candidate object trajectories based on a neural network cost function of a combination of a grasp quality cost and a task-specific cost by using gradient descent to minimize the neural network cost function.
A computer-implemented method for grasp selection, comprising:
- receiving one or more candidate object trajectories and a current grasp of a robotic hand on an object;
- sampling random candidate grasps for the one or more candidate object trajectories based on the current grasp;
- generating one or more grasps to be optimized for each of the one or more candidate object trajectories based on the sampled candidate grasps;
- and
- optimizing one or more of the grasps to be optimized for each of the one or more candidate object trajectories based on a neural network cost function of a combination of a grasp quality cost and a task-specific cost by using gradient descent to minimize the neural network cost function.
Dependencies are inferred from the claim text.