Robot Grasp Point Selection Using Depth-Based Tactile Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current robot grasping operations are inefficient and potentially damaging due to the need for multiple contact attempts to gather tactile data, which can be time-consuming and result in object damage.
Innovation Solution
A method and system utilizing depth perception modeling, specifically a neural network trained by deep learning, to estimate tactile output from depth data, allowing a robot to determine a potential grasp point and grasp an object without physical contact, thereby reducing the need for repetitive attempts.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If robots use tactile data from repetitive contact attempts to learn object characteristics, then grasping accuracy is improved, but time consumption and object damage risk increase
Solution Approach 1:
The system performs preliminary action by using depth perception modeling to predict tactile output and identify optimal grasp points before the robot makes physical contact with the object. This allows the robot to plan its grasping strategy in advance, eliminating the need for multiple repetitive contact attempts and significantly reducing time consumption.
Solution Approach 2:
The patent replaces the mechanical tactile sensing system with an optical depth perception system. Instead of relying on physical contact and tactile sensors to gather object characteristics, the system uses depth images and neural network modeling to predict tactile output, substituting mechanical interaction with optical field-based measurement.
2Loss of information
If robots use repetitive contact attempts to generate tactile data, then object characteristics are identified, but object damage occurs
Solution Approach 1:
The patent replaces mechanical tactile sensing with optical depth perception and computational modeling. The neural network model predicts tactile output based on depth image data, allowing the system to identify object characteristics such as shape, material, and surface properties without physical contact, thereby eliminating object damage from repetitive attempts.
Solution Approach 2:
The system creates a virtual copy or model of the object's tactile properties through depth perception modeling. Instead of physically contacting the object to sense its characteristics, the neural network generates a predictive model of tactile output based on depth image data, allowing virtual exploration of object properties without physical interaction.
3Reliability
If multiple contact attempts are made to learn object characteristics, then grasping reliability is improved, but operational efficiency decreases
Solution Approach 1:
The system performs preliminary analysis using depth perception modeling to identify optimal grasp points and predict tactile output before execution. This preliminary action ensures high grasping reliability by pre-planning the best contact points based on predicted object characteristics, while eliminating repetitive attempts and improving operational efficiency.
Solution Approach 2:
The patent replaces mechanical trial-and-error tactile sensing with optical depth perception and neural network prediction. This substitution maintains grasping reliability by accurately predicting object characteristics and optimal grasp points, while dramatically improving productivity by eliminating multiple contact attempts.
Data Source
AI summary
Methods, grasping systems, and computer-readable mediums storing computer executable code for grasping an object are provided. In an example, a depth image of the object may be obtained by a grasping system. A potential grasp point of the object may be determined by the grasping system based on the depth image. A tactile output corresponding to the potential grasp point may be estimated by the grasping system based on data from the depth image. The grasping system may be controlled to grasp the object at the potential grasp point based on the estimated tactile output.