Robotic Item Handling With 3D Sensing for Grasp-or-Sweep Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual loading and unloading of vehicles at material handling sites is costly and inefficient, necessitating an improved system for quickly and effectively handling bulk quantities of stacked or scattered items.
Innovation Solution
A robotic item handler equipped with machine learning-based decision making, utilizing multiple sensors to determine the optimal operating mode between grasping and sweeping, adjusting weights based on model performance and past heuristics to efficiently handle items.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated loading/unloading systems are implemented, then productivity and cost efficiency are improved, but device complexity increases
Solution Approach 1:
The robotic item handler is divided into distinct functional modules: a robotic arm with end effector for grasping operations, a platform for sweeping operations, multiple sensor devices for detection, and a processing unit for decision-making. Each module operates semi-independently, allowing the system to achieve high productivity through coordinated automation while managing complexity through modular design.
Solution Approach 2:
The robotic item handler is designed with multi-functionality to handle both stacked and scattered items using two different operating modes (grasping and sweeping). The system can adaptively switch between modes based on item configuration, enabling a single device to perform multiple handling tasks that would otherwise require different specialized equipment, thus improving productivity without proportionally increasing device complexity.
2Adaptability or versatility
If machine learning-based decision making is used to select operating modes, then adaptability is improved, but device complexity increases
Solution Approach 1:
The system employs sensor devices that continuously detect item characteristics and provide feedback to the processing unit. The processing unit uses machine learning algorithms to analyze this feedback and dynamically select the appropriate operating mode (grasping or sweeping). This feedback loop enables the system to adapt to different item configurations automatically, improving versatility while the automated decision-making process manages control complexity by replacing manual intervention with algorithm-based selection.
Solution Approach 2:
The robotic item handler performs self-service through autonomous decision-making. The processing unit automatically evaluates sensor data, determines the optimal operating mode without human intervention, and controls the robotic arm or platform accordingly. This self-service capability improves adaptability to various item arrangements while reducing the complexity of external control systems by embedding the decision-making logic within the robot itself.
3Measurement precision
If multiple sensor devices and machine learning models are deployed, then measurement precision is improved, but use of energy increases
Solution Approach 1:
The system uses multiple sensor devices to detect item characteristics, but the machine learning model processes only the essential features needed for mode selection rather than analyzing all sensor data in full detail. This partial processing approach maintains sufficient measurement precision for distinguishing between stacked and scattered items while reducing the computational energy required compared to comprehensive analysis of all sensor inputs.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for controlling a robotic item handler is described. The method includes obtaining first point cloud data related to a first three-dimensional (3D) image and second point cloud data related to a second 3D image, captured by a first sensor device and a second sensor device respectively. Further, the method can include transforming the first point cloud data and the second point cloud data to combined point cloud data that is used as an input to a convolutional neural network, to construct a machine learning model. The machine learning model can output a decision classification indicative of a first probability associated with a first operating mode and a second probability associated with a second operating mode. Furthermore, the method can include operating the robotic item handler according to the first operating mode or the second operating mode based on a comparison of the first probability and the second probability.