Back to results
US 2023/0080768 A1Published application
Object manipulation
AssigneeHonda Motor Co., Ltd.
InventorsRana SOLTANI ZARRIN, Katsu YAMANE
Life of this patent
Estimating the term…
Abstract
A robot for object manipulation may include sensors, a robot appendage, actuators configured to drive joints of the robot appendage, a planner, and a controller. Object path planning may include determining poses. Object trajectory optimization may include assigning a set of timestamps to the poses, optimizing a cost function based on an inverse kinematic (IK) error, a difference between an estimated required wrench and an actual wrench, and a grasp efficiency, and generating a reference object trajectory based on the optimized cost function. Grasp sequence planning may be model-based or deep reinforcement learning (DRL) policy based. The controller may implement the reference object trajectory and the grasp sequence via the robot appendage and actuators.
Claims20
A system for object manipulation, comprising:
- a planner performing:
- object path planning to determine a first object pose, a second object pose, and an Nth object pose;
- object trajectory optimization by:
- assigning a set of timestamps to the first object pose, the second object pose, and the Nth object pose;
- optimizing a cost function based on a first factor indicative of an inverse kinematic (IK) error, a second factor indicative of a difference between an estimated required wrench and an actual wrench, and a third factor indicative of a grasp efficiency;
- and
- generating a reference object trajectory based on the optimized cost function;
- and
- grasp sequence planning generating a grasp sequence by a first type of grasp sequence planning or a second type of grasp sequence planning,
- wherein the first type of grasp sequence planning is based on the optimized cost function and minimizing added contact points and/or removed contact points,
- wherein the second type of grasp sequence planning is based on a deep reinforcement learning (DRL) policy trained based on a reward function, wherein the reward function is based on a first term indicative of the IK error, a second term indicative of the difference between the estimated required wrench and the actual wrench, and a third term indicative of the grasp efficiency;
- and
- a controller, implementing the reference object trajectory and the grasp sequence via a robot appendage and one or more actuators.
10 more elements
A method for object manipulation, comprising:
- performing object path planning to determine a first object pose, a second object pose, and an Nth object pose;
- performing object trajectory optimization by:
- assigning a set of timestamps to the first object pose, the second object pose, and the Nth object pose;
- optimizing a cost function based on a first factor indicative of an inverse kinematic (IK) error, a second factor indicative of a difference between an estimated required wrench and an actual wrench, and a third factor indicative of a grasp efficiency;
- and
- generating a reference object trajectory based on the optimized cost function;
- performing grasp sequence planning generating a grasp sequence based on a first type of grasp sequence planning or a second type of grasp sequence planning,
- wherein the first type of grasp sequence planning is based on the optimized cost function and minimizing added contact points and/or removed contact points,
- wherein the second type of grasp sequence planning is based on a deep reinforcement learning (DRL) policy trained based on a reward function, wherein the reward function is based on a first term indicative of the IK error, a second term indicative of the difference between the estimated required wrench and the actual wrench, and a third term indicative of the grasp efficiency;
- and
- implementing the reference object trajectory and the grasp sequence via a robot appendage and one or more actuators.
8 more elements
A robot for object manipulation, comprising:
- a robot appendage;
- one or more actuators configured to drive one or more joints of the robot appendage;
- a planner performing:
- object path planning to determine a first object pose, a second object pose, and an Nth object pose;
- object trajectory optimization by:
- assigning a set of timestamps to the first object pose, the second object pose, and the Nth object pose;
- optimizing a cost function based on a first factor indicative of an inverse kinematic (IK) error, a second factor indicative of a difference between an estimated required wrench and an actual wrench, and a third factor indicative of a grasp efficiency;
- and
- generating a reference object trajectory based on the optimized cost function;
- and
- grasp sequence planning generating a grasp sequence by a first type of grasp sequence planning or a second type of grasp sequence planning,
- wherein the first type of grasp sequence planning is based on the optimized cost function and minimizing added contact points and/or removed contact points,
- wherein the second type of grasp sequence planning is based on a deep reinforcement learning (DRL) policy trained based on a reward function, wherein the reward function is based on a first term indicative of the IK error, a second term indicative of the difference between the estimated required wrench and the actual wrench, and a third term indicative of the grasp efficiency;
- and
- a controller, implementing the reference object trajectory and the grasp sequence via the robot appendage and one or more of the actuators.
12 more elements
Dependencies are inferred from the claim text.