Operation planning system for robot

• US 5,513,299 A
• Filed: 01/13/1994
• Issued: 04/30/1996
• Est. Priority Date: 10/31/1990
• Status: Expired due to Fees

• Current Assignee: Sanyo Electric Company Limited (Panasonic Holdings Corporation)
• Original Assignee: Sanyo Electric Company Limited (Panasonic Holdings Corporation)
• Inventors: Takahashi, Hironobu, Terasaki, Hajime
• Primary Examiner(s): Downs, Robert W.
• Application Number:
  US08/181,097
• Time in Patent Office: 838 Days
• Field of Search: 395/90, 395/97
• US Class Current: 700/255
• CPC Class Codes: B25J 9/1661 characterised by task planning, object-oriented languages
• CPC Class Codes: B25J 9/1669 characterised by special application, e.g. multi-arm co-operation, assembly, grasping
• CPC Class Codes: G05B 2219/39082 Collision, real time collision avoidance
• CPC Class Codes: G05B 2219/39476 Orient hand relative to object
• CPC Class Codes: G05B 2219/39479 Grip, release again to put object in correct position in tray, regrip and move
• CPC Class Codes: G05B 2219/39536 Planning of hand motion, grasping
• CPC Class Codes: G05B 2219/40437 Local, directly search robot workspace
• CPC Class Codes: G05B 2219/49145 Spheres replace object, check first collision for large spheres, then small

Abstract

Pick-and-place grasp candidates for positions at which an object can be picked and placed on a table are calculated based on the initial and goal states of the object, the surrounding environment state at the initial and goal states and the geometrical data of a robot hand. A space characterization is prepared which represents various states relating to the object around the grasp position in each of the calculated pick-and-place grasp candidates. The magnitude of the space in which the hand can be moved is estimated from the prepared space characterizations. Based on the results, position and orientation for pick-and-place grasp is selected from the pick-and-place grasp candidates. If the object cannot be moved from its initial state to its goal state through one grasp and movement action, the object is once returned onto the table and regrasped by the robot hand. Regrasp orientation of the object is selected from stable orientations which depend on the geometric data of objects to be handled. In such a manner, a robot operation plan can be prepared automatically.

Claims

1. A robot operation planning system for planning the grasp and movement of an object by a hand of a robot, said system comprising:

means providing data representing a three-dimensional model of the environment in which the object is disposed, including any obstacles surrounding the object in the environment, geometric data representing the shape of the object, and data representing an initial state and a goal state of the object on a table;

stable position calculating means connected to receive the geometric data for determining, from the data, all possible stable orientations in which the object can be stably placed on the table, and calculating intermediate locations of the object in the stable orientation during movement from the initial state to the goal state such that the robot hand can operate at the intermediate states without being affected by obstacles on the table, said stable position calculating means providing data representing the calculated intermediate states;

state transition path searching means connected to receive data representing the model for searching for a state transition path where the object is transferred from the initial state to the goal state and via at least one of the intermediate states if necessary; and

regrasp location determining means connected to receive the data representing the calculated intermediate states for determining a location for regrasping the object when the object is not directly moved from the initial state to the goal state, and is temporarily placed and regrasped at an intermediate states, the regrasp location being determined by preferentially searching for a location on the table which has a maximum distance from surrounding obstacles thereon.

2. A robot operation planning system as defined in claim 1, further comprising:

grasp candidate calculating means for calculating candidates for grasping position and orientation at which an object can be grasped at a given grasp position, at least based on the geometric data of said object and the robot hand;

open space pyramid selecting means for dividing the space around each grasp candidate into space pyramids each having an apex corresponding to the grasp position and a predetermined magnitude and for selecting, as open space pyramids, space pyramids having no interference with surrounding obstacles from said divided space pyramids;

distance transforming means for determining the distance between each of the selected open space pyramids and a non-open space pyramid; and

grasp selecting means for determining the grasp position and orientation of the robot hand, based on both the distance of the respective open space pyramids determined by said distance transforming means and the grasp candidates calculated by said grasp candidate calculating means, whereby an object grasp can be determined when the object is shifted form one state to another state.

3. A robot operation planning system as defined in claim 2 wherein said grasp selecting means selects a grasp candidate which has the maximum distance value.

4. A robot operation planning system as defined in claim 3 wherein said grasp candidate calculating means recognizes, in consideration of the rotational symmetry of the object, a plurality of grasps from each pair of opposite directions in the rotational symmetry to be the same grasp candidate.

5. A robot operation planning system as defined in claim 2 wherein said grasp candidate calculating means calculates a grasp candidate which can pick and place the object on the table, based on the geometric data of object, the hand of the robot and environment in one state, and another state to be shifted to from said one state.

6. A robot operation planning system as defined in claim 5 wherein said grasp candidate calculating means determines a zone by projecting a region of the object where the robot hand can be grasped at the initial and goal states, and regions of the obstacles at the initial and goal states, on to the gripping plane.

7. A robot operation planning system as defined in claim 6 wherein the robot hand includes two fingers each having a finger tip with a circular portion for contacting the object.

8. A robot planning system as defined in claim 1 in combination with the robot and means for controlling the robot hand in response to determinations made by said regrasp location determining means.