Robotics and Automation Expert
page 1 page 3 page 4 page 5 page 6

Persistent Memory Improves Robotic System & Machine Safety (page 2)

Persistent memory’s key essential feature is a memory write-protected interlock system. Many microprocessors have protected memory spaces whereby a program is stopped when it attempts to write into another program’s data or program area. Though this is typically found in PC-class processors (Pentium, Power Architecture, and 68000 for example), most DSP processors used in real-time motion control do not have this capability. A simple write protection scheme involves using an I/O port pin to control write protection in the memory access controller. Non-volatile memory updates then require the write protect interlock to enable access. This protects the nonvolatile data from inadvertent write operations. Another issue is battery switchover and power-off write protection. Typically very fast power spikes can disrupt operations before the analog comparator circuits can act, so it is important to ensure that the proper memory protection is in place.
Many robotic systems exist as a part of a larger coordinated system and the complexity is growing each year. In this environment, an unrecoverable fault can have ripple effects throughout the operation. Power grids are becoming more and more overloaded as energy use grows, thus placing an increasing strain on the power transmission infrastructure. Fast, nonvolatile memory throughout the system is advised so that the entire operation can be restored completely and as fast as possible with minimal scrap, rework, line downtime, and most importantly, the safety of operators and other manufacturing personnel.

Successful coordinated control structure can be enhanced through transaction logging. Transaction logging is a two-step process. First, record the robotic event that needs to be completed into the log. Second, when the event is completed, reconcile the log. This is much like reconciling a checkbook, which ensures that the bank’s records and customer’s records are in sync. This concept also applies to a multitude of robotic coordinated tasks.

For example, in a contour profile (a description of the robot’s arm movement) a piecewise linear path is constructed of the arm’s tool point. Inverse kinematics translates the tool path into angle and velocity changes for each axis for coordinated motion. A smaller contour interval results in a better approximation of the exact curve. At the end of the profile interval, each axis should be at the desired location. However, coordinated control requires a position check to ensure that each axis has completed the operation and is close enough to the ending coordinate to be within acceptable limits. The end result is a faster control loop that provides smooth curve fit quality and highest throughput. Having a transaction logging process capable of high-speed operation is therefore imperative.

The preceding illustration is a simple example of coordinated axis control with robust transaction logging. However, far more complex processes involving conveyors, multiple robots, handlers, and upstream feed information require the global transaction logging processes to ensure proper coordination simultaneously at many levels. The key requirement of a transaction log is to ensure that actions or events have actually occurred in the right order or in the logical arrangements the applications requires. This coordination method requires nonvolatile storage of the rapidly changing parameters. Central logging over a network is not viable because of the risk associated with overloading the network’s bandwidth and possible losing valuable data.