Robot Grasp Planning Using Multi-Angle Imaging in Motion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing robot systems that stop to acquire information about a target object for grasping face challenges in recognizing the grasping position accurately, leading to a low success rate and increased time requirements due to limited directional information acquisition.
Innovation Solution
A robot equipped with an image-pickup acquisition unit that captures images from multiple directions while moving towards the target, using a learned model to calculate and select the most likely grasping position, thereby reducing unobservable parts and optimizing grasping efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the robot stops to acquire information on the target object, then the information acquisition is performed from one direction, but the grasping position cannot be recognized appropriately and the grasping time increases
Solution Approach 1:
The robot transitions from a static stopping approach to a dynamic moving approach, acquiring images continuously while approaching the target object. This dynamic image acquisition allows multiple directional views to be captured during motion, improving grasping position recognition without increasing total grasping time.
Solution Approach 2:
The system adds the time dimension to image acquisition by capturing images continuously while the robot moves, rather than stopping at a single position. This transforms the acquisition process from a single-point static capture to a multi-point dynamic capture along the approach trajectory, enabling comprehensive view acquisition.
2Loss of information
If the robot stops to acquire information on the target object, then the information is acquired from one direction, but the number of unobservable parts increases
Solution Approach 1:
The robot performs image acquisition dynamically during movement rather than statically when stopped. This allows the imaging device to capture the target object from multiple directions along the approach path, reducing unobservable parts while maintaining efficient time utilization.
Solution Approach 2:
The image acquisition process continues uninterrupted during the robot's approach movement. Instead of stopping to acquire images and then resuming movement, the system performs both acquisition and movement simultaneously, ensuring continuous useful action and reducing total task time.
3Reliability
If the robot stops when approaching the target object, then the information acquisition is performed, but the grasping success rate decreases
Solution Approach 1:
The system implements dynamic image acquisition during the approach phase, allowing the robot to gather comprehensive visual information from multiple directions while moving toward the target. This improves grasping success rate by providing better spatial understanding without requiring additional stopping time.
Solution Approach 2:
The robot acquires image information preliminarily during the approach phase before actual grasping execution. By gathering comprehensive visual data while moving toward the target, the system prepares sufficient information for accurate grasping planning, improving success rate without extending the overall grasping timeline.
Data Source
AI summary
A robot including a manipulator includes: an image-pickup acquisition unit configured to acquire an image of an environmental space including a target object to be grasped; and a control unit configured to control a motion performed by the robot, in which the control unit causes the robot to acquire, by the image-pickup acquisition unit, a plurality of information pieces of the target object to be grasped while it moves the robot so that the robot approaches the target object to be grasped, calculates, for each of the information pieces, a grasping position of the target object to be grasped and an index of certainty of the grasping position by using a learned model, and attempts to grasp, by moving the manipulator, the target object to be grasped at a grasping position selected based on a result of the calculation.