Multiple object grasping using robotics
Systems and methods for multiple object transferring using robotics. A system includes a robotic hand with a base and fingers capable of grasping objects. A robotic arm is coupled to the robotic hand and is capable of moving the robotic hand. One or more circuits are configured to operate the robotic hand and the robotic arm by identifying a pre-grasp configuration for the robotic hand and executing a transfer routine based on a Markov decision process model to operate the robotic hand and the robotic arm such that they move multiple objects from a first location to a second location.
1 . A system, comprising:
a robotic hand comprising a base and fingers, the robotic hand capable of grasping objects; a robotic arm coupled to the robotic hand and capable of moving the robotic hand;
one or more circuits for operating the robotic hand and the robotic arm, the one or more circuits configured to:
identify a pre-grasp configuration for the robotic hand based on a target quantity of objects to be grasped by the robotic hand;
operate the fingers of the robotic hand m accordance with the pre-grasp configuration; and
execute a transfer routine based on a Markov decision process to operate the robotic arm and the robotic hand such that the robotic hand grasps the target quantity of objects in a first location and the robotic hand and the robotic arm transfer the target quantity of objects to a second location, wherein numbers of objects in the second location are modeled as states in the Markov decision process.
2 . The system of claim 1 , wherein the target quantity is greater than one of the objects.
3 . The system of claim 1 , wherein the one or more circuits are configured to identify the pre-grasp configuration for the robotic hand based on a spread angle associated with at least one of the fingers of the robotic hand.
4 . The system of claim 3 , wherein the one or more circuits are configured to identify the pre-grasp configuration for the robotic hand based on a probability of successfully grasping the target quantity of objects associated with the spread angle.
5 . The system of claim 1 , wherein the one or more circuits are configured to identify the pre-grasp configuration for the robotic hand based on a weighted sum of probabilities associated with different quantities of objects capable of being grasped by the robotic hand.
6 . The system of claim 1 , wherein the one or more circuits are configured to identify the pre-grasp configuration for the robotic hand based on a volume associated with the pre-grasp configuration.
7 . The system of claim 1 , wherein the Markov decision process models grasping, lifting, and transferring as performed by the robotic hand and the robotic arm as actions within the Markov decision process.
8 . A method, comprising:
identifying a pre-grasp configuration for a robotic hand comprising fingers based on a target quantity of objects to be grasped by the robotic hand;
operating the fingers of the robotic hand in accordance with a spread angle of one or more of the fingers of the robotic hand associated with the pre-grasp configuration; and
executing a transfer routine based on a Markov decision process to operate a robotic arm coupled to the robotic hand such that the robotic hand grasps the target quantity of objects in a first location and transfers the target quantity of objects to a second location, wherein numbers of objects in the second location are modeled as states in the Markov decision process.
9 . The method of claim 8 , wherein identifying the pre-grasp configuration for the robotic hand comprises identifying the pre-grasp configuration based on a probability of successfully grasping the target quantity of objects associated with the spread angle.
10 . The method of claim 8 , wherein identifying the pre-grasp configuration for the robotic hand comprises identifying the pre-grasp configuration based on a weighted sum of probabilities associated with different quantities of objects capable of being grasped by the robotic hand.
11 . The method of claim 8 , wherein identifying the pre-grasp configuration for the robotic hand comprises identifying the pre-grasp configuration based on a volume associated with the pre-grasp configuration.
12 . The method of claim 9 , wherein executing the transfer routine based on the Markov decision process comprises executing the transfer routine based on the Markov decision process wherein the Markov decision process models different grasping actions performed by the robotic hand as actions within the Markov decision process.
13 . The method of claim 8 , further comprising:
identifying an end-grasp configuration for the robotic hand based on the target quantity of objects to be grasped by the robotic hand; and
operating the fingers of the robotic hand in accordance with the end-grasp configuration to perform a grasping action using the robotic hand.
14 . A method, comprising:
identifying a pre-grasp configuration for a robotic hand comprising fingers based on a target quantity of objects of a collection of objects to be grasped by the robotic hand;
identifying an end-grasp configuration for the robotic hand based on the target quantity of objects to be grasped by the robotic hand;
operating the fingers of the robotic hand in accordance with the pre-grasp configuration; operating a robotic arm coupled to the robotic hand to move the robotic hand near the
collection of objects;
operating the fingers of the robotic hand in accordance with the end-grasp configuration such that the robotic hand grasps the target quantity of objects; and
executing a transfer routine based on a Markov decision process to operate the robotic arm and the robotic hand such that the robotic arm and the robotic hand transfer the target quantity of objects from a first location to a second location, wherein numbers of objects in the second location are modeled as states in the Markov decision process.
15 . The method of claim 14 , wherein identifying the pre-grasp configuration for the robotic hand comprises identifying the pre-grasp configuration based on a spread angle associated with at least one of the fingers of the robotic hand.
16 . The method of claim 15 , wherein identifying the pre-grasp configuration for the robotic hand comprises identifying the pre-grasp configuration based on a probability of successfully grasping the target quantity of objects associated with the spread angle.
17 . The method of claim 14 , wherein identifying the pre-grasp configuration for the robotic hand comprises identifying the pre-grasp configuration based on a volume associated with the pre-grasp configuration.