The automated bin picking industry uses autonomous robotic arms technology to facilitate work.
Thanks to automation, bin-picking can be conducted in semi-structured surroundings without human intervention.
The process is considered a cost-effective commercial application because of its significant return on investment (ROI) and less need for human labor. Read on to learn more about automated bin-picking.
Automated Bin-Picking
Automated bin picking is collecting known objects with different poses from a given place for various tasks using robots and automation.
Some good examples of bin-picking include collecting machine components for assembly, assembling things for machine tending, and arranging products on shelves.
The processes are primarily operated by the 3D vision and smart manipulation methods.
Automated bin-picking uses computer vision to identify objects and robotics to control the robot during collection.
Automation bin-picking uses perception modules to observe the environment, identify the objects that need picking from a given bin and then decide the appropriate approach to collect and put them at a designated location.
The robot then picks the object and places it in the identified location using various motion-planning strategies.
The Automated Bin-Picking Pipeline
Visual perception and robot motion are the leading categories of the automated bin-picking process.
The former has sub-categories used to identify items and create the clutch pose. The latter has control algorithms that facilitate the transfer and drop process.
Items In The Bin
The success of automated bin-picking is dependent on the type of objects selected for picking.
The success is high when picking objects with symmetrical shapes, bright colors, and other significant features that are easy to spot.
Objects with varying heights, distorting properties, and geometric shapes are difficult to observe and their bin-picking process is less successful.
There are three bin-picking categories: structured bin-picking, semi-structured bin-picking, and unstructured bin-picking, depending on how objects are put in the bin.
Structured bin-picking has a predictable arrangement, semi-structured bin-picking has minor inaccuracies, while unstructured bin-picking involves random object placement.
The structured and semi-structured categories don’t require visual sensing are not common in industrial settings and require more work than beats automation purpose.
Visual Perception For Automated Bin-Picking
Automated bin-picking uses computer vision and deep learning to operate. Here are the visual perception module components applied in automated bin-picking.
1. Object Detection – the most challenging part of automated bin-picking is identifying individual objects. Object detection uses computer vision methods for identifying simple objects, and deep learning techniques for identifying complex objects.
2. Object Recognition – this component comes into play after an object has been detected. It depends on how well an object was picked. Camouflaged and occluded objects are the most difficult to recognize.
3. Object Pose Estimation – Once an object is recognized, a software CAD model and other localization algorithms are used to identify its pose on the bin. 3D vision is used in this scenario to ensure accuracy.
4. Grasp Pose Estimation – After object pose estimation, grasp pose estimation is done to pick the object. This step uses the robot’s EOAT geometric model to pick the object.
These steps are applicable in a static bin. The object pick-up involves these four steps and visual serving in the case of a moving bin or a conveyor belt.
Robot Motion In Bin-Picking
After the object selected for picking is identified and a grasp pose is formed, the robot moves to the target to conduct the pick-up.
Conventional motion planning or visual serving are the commonly used pick-up point approaches used during robot motion in automated bin picking.
EOATs possess torque sensors to ensure successful object grasp, which starts the drop-off process. Collisions can deter successful grasps.
A comprehensive solution can effectively resolve these issues and complete a successful pick-up.
EOAT For Automated Bin-Picking
EOAT is essential in successfully implementing automated bin-picking. Traditional two-fingered grippers work well for rigid objects.
However, they are not suitable for larger and deformable objects. Suction-based grippers are being implemented to handle deformable and extensive objects.
