It's a compelling concept: Use an inexpensive PC to control the axes of a robot but do so in a way that avoids the cost of a Windows operating system.
This is the model Automated Assemblies Corp. followed in fielding a new robot for the injection-molding industry.
No question PC-based controls have become more widely used in motion control, but "We'd been leery of them because of Windows," says AAC Product Development Manager David Lee. ACC specifically didn't like the licensing fees involved or the need to stay current with new versions of the operating system. The latter point could involve significant effort over the average 10-yr life of an industrial robot.
The way around this dilemma was through an approach that used the Linux open source operating system. For its new Raptor Series robots, ACC handles motion control and work-cell functions through a PC-based platform from Motion Engineering Inc., Santa Barbara, Calif. The real-time version of Linux running on the MEI board "gives us more control of our destiny," explains Lee. "We can customize the kernel as necessary and use only the parts of the software we need. Plus relying on an outside supplier for motion control lets us be experts in building robots and not in building motion platforms."
Off-loading motion control to a card also let ACC use a less powerful PC to handle the balance of chores for the work cell. Besides serving as an operator terminal, the PC permits interactions with I/O through the same interface that handles servo axes. ACC also uses this capability to give custom molders a way to quickly change their production setup, thus shortening turnaround time for new jobs. "When a mold is open, it is considered downtime," explains Steve Braig, ACC president. "The goal is to have the mold always full and making product, so we want to reduce the time molds are open each cycle. One solution is to make robots go faster."
An XMP SynqNet-PCI network and controller handles all servo axes and I/Os. XMP controllers incorporate digital signal processors to manage dedicated motion. A map architecture works over the computer PCI bus for handling tasks between host and controller. Motion programming takes place via MPI, an ANSI-compatible object-oriented C/C++ motion-programming library that works with other SynqNet controllers. (Other programming environments include ActiveX and MatLab/Simulink, which are accessed via plug-ins.) MPI is platform-independent. In addition, it supports CANOpen, so one programming interface handles motion and CAN I/O. The controllers can network via daisy chain and ring topologies. Ring topologies let "self-healing" fault tolerant operation work around cable faults between nodes, motion continues via duplex communication from the controller.
The SynqNet network is based on IEEE 802.3. Only one cable runs between controller and drives, so wiring and debugging could be done on a prototype in a matter of hours. And the small size of the controller lets it sit close to motors, further reducing the amount of cabling. The network handles torque-update rates of 48 kHz, downloads configuration files and firmware, and makes it possible to monitor cells remotely via Ethernet.
SynqNet supports up to 32 tightly synchronized high-performance axes, keying on applications with end-of-tool servo requirements. AAC developed its own graphic user interface for a drag-and-drop touchscreen to control and teach the robot routines, as well as other equipment in work cells such as vision-inspection equipment, and conveyors. AAC also developed routines, which let the controller automatically adjust speed and motion trajectories based on type and weight of parts being taken from the molds.
MEI provided several software aids for programming motion, including MotionScope, which graphs and analyzes real-time data. Lee used Linux utilities to let a remote host running MotionScope analyze motion data fed back over standard Ethernet connections. "This was invaluable for quickly analyzing and improving motion performance."
With the success rollout of the new robots, Steve Braig, AAC president, sees a greater role for automation in the plastics industry that will help maximize efficiency. "Once the robot grasps the parts, you have the option of inspecting them on-the-fly with smart cameras tied directly to the PC. Parts can then go directly into sorting trays and be further packaged as part of a complete work-cell solution.