Grab the Future
Bin picking with a 3D Sensor
Isra Vision offers a highly flexible and cost effective solution for bin picking with its new Shapescan 3D sensor. Random parts can be taken directly from the transport container by the robot and - without complex mechanical equipment - be entered into the manufacturing process.
Transport packaging is a necessary part of assembly technology. For example, it is necessary to take parts continuously from containers or pallets and introduce them to the manufacturing line. Up to now, manufacturing operations have had to make do with manual part delivery, which is associated with high logistical costs. In most cases, the parts are delivered ready for sorting, but are not actually sorted. Another alternative involves the operator having to manually remove the parts. For the most part, mechanical solutions have been employed. For example, shaker conveyors or vibratory feeders can sort small parts. Mechanical sorters take over the supplying and of parts, but are a costly and sometimes unreliable solution.
Direct access to the containers, also called bin picking or random bin picking, offers the advantages of saving space and reducing costs in comparison to current methods, and requires significantly fewer mechanical components. Researchers have worked on this problem since the 1980's. Up to now, many custom approaches have been tried with no concrete results. Robot based isolation techniques for random access to complex objects have always had a weak point due to problems with image processing. In practice, the robot does not always execute the same movement; it must be capable of responding flexibly to different object locations. Previous solutions with 2D camera technology offer only limited solutions. These systems often fail due to difficult conditions such as low contrast and not being able to see depth. Bright parts on a light background, for example, cannot be differentiated. These systems have therefore not been successful in the market.
The Solution: a New 3D Sensor
Isra Vision is presenting the Shapescan 3D sensor which achieves perfect automated control within the container. This sensor makes it possible to exactly determine the position of individual parts in a container using two standard cameras and special laser lighting.
The sensor is stationary, positioned over the box, and is able to view a bit more than the size of a standard European pallet in one view. Complex Kinematics like feed units, are not necessary for the sensors to carry out their tasks. The system nevertheless works very quickly: The scan time for a typical container (about 1,200 mm x 1,000 mm x 800 mm) requires less than two seconds for the entire volume. Shapescan provides a precision of a few millimeters in this short period of time. If other container sizes are used, the distance between the two sensor camera units can be adjusted. The sensors can also be attached in a mobile fashion on a robot. In this case, the robots see only individual sections of the container rather than the whole area in one view. The sensor can construct a complete view of the area from various visual sections. In addition, there is a sensor version available with four cameras which are designed for applications which require high speed.
New Methods: 3D Shape Sampling
The special intelligence provided by Isra's sensor solution resides in its sophisticated software: The 3D cameras scan the containers and create a so-called "3D cloud," which represents the objects captured in the transport packaging. The new assessment methods developed for this 3D shape sampling are first determined in a geometrically simple form in the point cloud. The software then puts together complex objects from these simple representations. It can identify different shaped objects without prior knowledge. It assures robust recognition of parts geometry, even for unsorted objects. In order to determine the position of the objects in depth, a laser line projection is used. This makes it possible for the sensors to see in the third dimension.
Since several parallel lines are evaluated at the same time, the sensor works much quicker than other sensors. The "depth view" makes it possible for robots to completely empty the container without intermediate placing of the objects. Access is no longer limited to a specific level. In addition, this process is not sensitive to contamination.
Reduce Costs, Increase Flexibility
Isra Vision's 3D Shapescan sensor presents an easily handled and highly flexible solution for automated access into containers for simple to highly complex parts, even in rough production environments and at high process speeds. Production and assembly operations can benefit with these sensor systems by saving significant costs, improve manufacturing throughput and increase flexibility of production in a significant manner, all while producing low noise levels. The systems are simple to install and require very little additional space.