3D image processing: More flexibility and performance through programmability

Many of SICK’s sensors and cameras are now open and programmable as part of the SICK AppSpace ecosystem. This means that integrators and OEMs have the freedom to integrate their applications and ideas into the devices, explains Nina Hammerin, 3D Vision Product Manager at SICK IVP.

The new TriSpectorP1000 programmable 3D camera is part of the SICK AppSpace development environment and opens up new possibilities to develop solutions utilising the values of 3D vision. The TriSpectorP1000 offers full flexibility for detection, as well as 2D and 3D inspection, of moving objects, thereby allowing tailored solutions for a wide range of image processing applications.

The TriSpectorP1000 is a compact, programmable 3D camera; its design combines laser illumination and camera optics in a single housing, creating a geometrically stable and identical measurement set-up in every application, and convenient integration of the sensor into the machine environment. There are three different field-of-view variants, making it possible to inspect objects from a few millimetres up to half a meter, thus enabling optimal implementation of the task. Image processing and analysis are integrated into the TriSpectorP1000 design. The high-resolution shape data measured by the factory-calibrated 3D camera is given millimetre values, and the processing results can be communicated directly from the device.

The possibility to program the TriSpectorP1000 in the SICK AppSpace development environment enables machine builders and integrators to implement tailor-made 3D vision solutions, including a web-based operator interface. For every individual application, the software modules of SICK AppSpace enables users to develop perfect-fit 3D image processing scripts and user interfaces, and to share these with their customers while protecting their intellectual property.

Programmability and SensorApp sharing initiate a paradigm shift

The programmability of optoelectronic sensors, RFID systems and cameras such as the TriSpectorP1000 has initiated a paradigm shift at many OEMs and integrators. Countless end customers now want solutions customised individually to their particular needs and limiting conditions, including their individual interfaces.

In many cases, the task is “only” to adjust the last few percentage points of the sensor’s functionality. In an image processing solution, this can, for example, be a so-called individualised SensorApp. Ideally, such SensorApps are shared via a platform to allow for broader use of new application solutions. The TriSpectorP1000 and other programmable sensors from SICK provide this possibility: Via the safe SICK AppPool cloud service, applications created by SICK and the integration partners of the SICK AppSpace community can be shared as SensorApps – the programming platform becomes a marketplace with secure data. Such SensorApps can also be switched very easily for changing the task of a sensor. For instance, the TriSpectorP1000 3D camera can be converted from performing a completeness check to, e.g. verify profiles or reading codes or characters by loading the right SensorApp.

TriSpectorP1000: Prepared for challenging 3D inspection tasks

Whether it is the high variance of parts, demanding object properties, fast processes, adverse ambient conditions, or stringent requirements for measurement accuracy, the programmable TriSpectorP1000 is the ideal choice for meeting even very complex requirements of 3D inspection in an uncomplicated and individualised manner. In terms of hardware, this 3D camera is designed to be consistent with stand-alone industrial usage: It has a space-saving and mechanically rugged anodised aluminium housing with enclosure rating IP 67 and is available with front screens made from either glass or plastic suitable for use with food.

For detection as well as 3D, 2D and profile inspection, the TriSpectorP1000 acquires real shape data. This data is independent of the colour, contrast and surface properties of the objects as well as background influences, and even factors related to ambient light. Results, for example picking coordinates, product volume and dimensions or the verified presence of a component, can be output via digital outputs and via the Gigabit Ethernet interface. Image processing scripts can be programmed with the functions in the SICK Algorithm API or with the HALCON library – a corresponding runtime license comes with the TriSpectorP1000 3D camera.

Inline inspection – flexible and tailor-made

The TriSpectorP1000 can be used as an inline inspection system in a wide range of industries and applications, including volume and dimension measurement of objects, visual guidance of pick-and-place robots, object counting and position checks, contour checks and profile verification, and anywhere efficient inspection of 3D object geometries is required. Stamped or engraved OCR characters can also be read and analysed.

For example, glue bead inspection during assembly of the front, rear and fixed side panes, body and gap size verification on the vehicle frame, and monitoring of the mechanical processing of cylinder heads are typical applications in automobile production. With the TriSpectorP1000, features such as differences in the thickness of the applied glue bead material, air pockets and gaps are reliably detected during glue bead inspection, even if the glue material colour changes. If specific SensorApps optimise different scenarios of the glueing process, they can be stored in the AppPool, and downloaded from there to the camera and be exchanged as often as desired.

Body and gap size verification also profit from the app sharing option, as assembly cut-outs, drill holes, and clearances can be inspected with the TriSpectorP1000. The diverse inspection tasks which are done on different types of vehicles from various viewing positions are programmed as individually optimised sensor apps and loaded directly into the 3D camera as appropriate for the application. The inspection of processed cylinders combines the glue bead and verification applications described; The 3D camera monitors the cylinder head for dimensional tolerance as well as checks the dispensed sealant to determine if the valve cover is assembled correctly.

In the food and packaging industry, the TriSpectorP1000 also proves its flexibility and versatility. In chocolate production, the programmable 3D camera monitors the product height in the casting moulds on the conveyor belt and detects over and underfilling with high precision, both with dark and white chocolate. In industrial bakeries and gastronomical production facilities, the TriSpectorP1000 monitors the volume, contour, shape and dimension of a wide range of baked goods and packaged products with toppings. These quality inspections ensure that only perfect products are packaged and delivered.

When gripping, sorting and packaging food with pick-and-place robots, the 3D camera detects the shape of each product and outputs the picking position, including the true height, as well as the rotation of every product. In this way, the robot can safely grab OK products and put them in primary packages while products that do not meet the picking criteria are left on the belt. At the same time, the camera can deliver quality information on the products, which makes it possible to inspect the manufacturing, transport and packaging processes and correct them if necessary.

The TriSpectorP1000 also handles the visual guidance of robots in the intralogistics sector. Objects on transport belts are reliably measured, and their positions determined. Using the same encoder to get input concerning the speed of the conveyor belt, the robot can now pick up and pass on every part with a highly secure grip and great precision.

In summary, the programmable TriSpectorP1000 3D camera enables a wide variety of tailored solutions for flexible automation, far beyond the examples mentioned above.