Anticipating every move
Does this sound familiar? Your team has designed a fast new machine, only to find that you cannot use it at anywhere near full speed. Well, here’s one fix.
Curtis S. Wilson
V.P. engineering,
research
Delta Tau Data Systems, Inc.
Power electronics, motor materials, and mechanical designs are making machines faster than ever. This single-board controller includes trajectory generation for full use of speed capabilities.
Digital motion control is
boosting output on everything
from high-end
machining centers to simple signmaking
devices. But often, these
systems aren’t used to their full dynamic
capacity, because tight corners
and sudden stops are too much
of a challenge to navigate safely,
within machine limits. So, designers
are sometimes forced to run entire
paths at a slow speed to prevent
machine damage. The situation
is analogous to having a powerful
sports car without headlights.
Newer look-ahead algorithms address this problem. They scout ahead in part programs, comparing dynamic requests from the program at a specified speed to the limitations of the machine. Then, if the program requests motion beyond machine capabilities, the look-ahead algorithm automatically substitutes a speed that the physical system can withstand.
But calculating a suitable speed is just the beginning. Look-ahead algorithms also calculate how a system will decelerate before and accelerate after a challenging maneuver. Systems cannot slow down immediately to a lower speed, so the algorithm works backwards along the path, to allow time to compute a profile before deceleration must actually begin. These algorithms can even bring systems to a complete stop from maximum speed without violating acceleration limits.
• As a required look-ahead time, this is maximum velocity divided by maximum acceleration.
• As a required look-ahead distance, it is the square of the maximum velocity divided by two times the maximum acceleration.
Quarter circle: With limiting
An X-Y path of a 90° arc move is approximated by 18 short linear moves blended together. Without look-ahead, all initial acceleration occurs in the first half of the first programmed move, covering only 2.5° — and deceleration occurs in the final 2.5°. Velocity transitions are much smoother after look-ahead acceleration limiting.
What’s more, look-ahead algorithms can check for several types of limits simultaneously. Checking for position (overtravel) limits may seem trivial, but by finding violations ahead of time and coming to a full stop before the limit is passed, rather than beginning to stop as the limit is being passed, can add several centimeters of active workspace around all edges of the machine.
In Cartesian systems, velocity limits mainly protect against programming mistakes, but in non-Cartesian systems programmed in tool-tip coordinates — such as a robot arm — look-ahead algorithms can protect against excessive velocities near singularities, where even a slow tool-tip move can require very high joint velocities.
Typically, though, the most important limits are those of acceleration. Look-ahead algorithms solve the biggest problem here — when required acceleration and especially deceleration cannot be accomplished in a single move block because of velocities that are high, and programmed move blocks that are short.
Want to use this article? Click here for options!
© 2008 Penton Media Inc.
advertisement







