3D for Better Control

If you can’t describe it you can’t control it.  And control is what its all about.  So we need to get better at describing what it is that our control system projects are going to do.

This has some serious implications.

First, if the programming language is not well suited to the task this will create a number of issues.  Ladder logic was never intended to deal with real time control. So blending Ladder logic with motion control, which can be the most demanding of real time control applications, doesn’t alway turn out well.  The precise coding of the control system processor now becomes an element of the control system behavior.  And often anomalies occur that are difficult to diagnose and which may be impossible to modify to achieve the desired results.

Maybe this is why embedded controls are gaining in popularity.  The computing power and connectivity of embedded controls has increased at least 10 fold in recent years making real time applications relatively straightforward.    But embedded controllers involve complex programming languages and require very exotic programming capability, often with expensive supporting “tool chains” are needed to develop and debug applications.  These costs make developing embedded applications suited to high volume products and not to one-of-a-kind machine control systems.

Control systems have become more interactive with data intensive applications like Excel Spreadsheet and higher level resource management applications.  So network connectivity and precise transaction capability with Windows applications becomes more appealing. It sounds more and more like a computer and not a control system.  And Windows isn’t really suited to the real time control or the hardware specific I/O that goes along with motion control applications.

PID as a control language for motion seems equally unsuited.  Its a great way to manage current and voltage relationships dynamically between motor and amplifier, but poorly equipped for managing the mechanical relationships between axes or mechanical properties like time and momentum.

In the mechanical realm of describing motion control, there is a similar problem.  Describing the mechanical task as a time-displacement trajectory seems an incredible understatement of the real work that needs to be done.  And there are no descriptive languages that do this well, at least, not yet.

But we are seeing the beginnings of a more comprehensive approach.  3D modeling software would seem to be the ideal environment from which a better description of the actual machine would be able to inform a control system program with valuable information about the performance of the actual mechanism.

Where could you get more perfect information about the changing nature of the reflected load of an actuator from a dependent axis as it creates changes in momentum of a primary axis.  As in a pick-and-place system.  The actual momentum as it changes over the entire motion profile can be extracted directly from the 3D model of the actuator and used as a filter to modify the commanded motion with incredible precision.  And this information can be used over and over, regardless of which location the actuator has to move to.

So we still don’t have the ideal solution.  But maybe that’s coming, soon.  The tools are there.

Universal Robotics Lauches 3D Software Compatible With Webcams

Universal Robotics, Inc., a software engineering company, announced the launch of two simple-to-use 3D vision software products: Spatial Vision and Spatial Vision Robotics. The products can turn any pair of webcams into a highly accurate, cost-efficient 3D vision system that can be employed in virtually any setting without expensive equipment.

With Spatial Vision and Spatial Vision Robotics, a user can plug in the cameras, calibrate their space and receive highly accurate measurements in under 30 minutes. These products will expand the use of 3D vision to markets where it hasn’t been feasible before.

3D vision systems offer many benefits over their 2D counterparts, including better accuracy and object identification and tracking, which are essential features in security, engineering and robotics applications from biometrics to real-time control of machines. Despite their benefits, broad adoption of 3D vision systems has been limited in many markets because the systems can be costly to implement and maintain.

Universal’s Spatial Vision products eliminate the need for the precision mounting, specialized cameras, and time-consuming set up that is required for many 3D vision systems. Using two webcams that can be set up and calibrated within a matter of minutes, Spatial Vision and Spatial Vision Robotics can determine the 3D position of any point relative to the cameras with millimeter accuracy.

The Spatial Vision product can be easily deployed in any setting in which cameras can be installed, including laboratories, office buildings, department stores and warehouses, and is an affordable solution for anyone looking for an accurate way to observe and measure an environment. It can be employed in security applications, measuring in-store foot traffic patterns, and more scientific applications requiring object tracking and visual analytics without a wand or sensing device. Spatial Vision offers 30 percent improved accuracy over 2D systems used in object identification and tracking applications, such as facial recognition and other biometrics. It is optimized for use with popular Logitech 9000 webcams, but can be  customized to work with any USB 2.0 camera.

Spatial Vision Robotics has been specially designed to be used in concert with automated machines. By adding LEDs to points of interest on moving machinery, Spatial Vision Robotics provides 3D position on the machine and its surroundings in robot coordinates as seen from the camera. The program enables 3D calibration between the extrinsic object of interest, the robot and the cameras, as well as intrinsic calibration with the cameras. It can work with any robot and is currently optimized for Yaskawa America (Motoman) SDA-series robots. Spatial Vision Robotics can be integrated with path planning and high-speed inverse kinematics to enable real-time control of robots.

Spatial Vision and Spatial Vision Robotics were created as part of the development of Universal’s signature technology, Neocortex™, a sensory-motor based form of artificial intelligence that enables moving machines to learn from their experiences and perform tasks that are unsafe or difficult for humans. Neocortex was developed over seven years with NASA and Vanderbilt University, and was funded by U.S. Department of Defense.

www.universalrobotics.com