DIRECTOR - TECHNICAL DEVELOPMENT / ENGINEERING / R&D
Technical leader and multi-disciplinary innovator with a strong background in systems, control theory and practice, robotics, C++, software algorithms and architectures, and distributed embedded systems. Ph.D. in Mathematics with expertise in the following areas:
* Real-Time and Adaptive Control Systems
* Object-Oriented Design / Implementation
* C++ Development
* Mathematical Analysis / Algorithm Design
* Technology Leadership and Team Building
* Concept Development / Innovation
* System Design / Redesign
* Distributed Processing / Control
SHAIN PARK CONSULTANTS, INC., Birmingham, MI 1997 - Present
Director of Technical Development / President
Formed Shain Park Consultants, Inc. to provide system design, software development, and consulting services to R&D groups across the U.S. I consulted in a variety of areas, from conveyer lines to railroad locomotives, and continued to research and develop control systems.
I helped clients to define their capabilities and needs, and to architect solutions. Where necessary, I invented extensions beyond what was commercially available. Solutions included: fault-tolerance, coherent data management, and motion control.
For several clients, under a variety of operating systems, such as QNX and MS Windows, I developed n-tier architectures with real-time capability for control. I also negotiated a strategic alliance with a drive manufacturer for motion controller development and sale.
RD INTERNATIONAL, LTD., Southfield, MI 1992 - 1997
Director of Technical Development / President and Co-Founder
Established RD International, Ltd. to research and develop new applications of control systems that would increase efficiency and productivity. I was approached by General Motors to apply my motion control expertise to improve the efficiency of a tandem press line. GM contributed the use of its laboratories and in-plant press line. I became technical leader of the project, with responsibility for team-building, technical innovation, and product development.
After studying the press line operation, I realized that improving coordination between machines would have the greatest impact on efficiency. Recognizing that coordination is limited both by forecasted system state uncertainty and by control specification language, I invented a new press control that both forecasted and reduced uncertainty in press position and I created a specialized, patented high level language for control and coordination. The language, which addressed general control system requirements (integrating I/O control, motion control and graphic interface control), incorporated a cross platform compatible compiler for high performance real-time implementation.
I also invented an advanced, patented, high performance motion control with new features for inter-device coordination. The new system controller addressed concerns common to all control systems: reliability, safety, time-critical response, run-time tuning, rapid restart after fault and diagnostics.
I was directly involved with the entire hardware/software design and development process for integrated line control, ultimately comprising over 1,000 code modules and 500,000 lines of C++. I developed an object-oriented architecture specification that allowed coding and debugging to be completed in less than one year.
In 1996, I formed and led an inter-corporate installation team in pilot tests on the GM tandem press line, controlling 6 robots and 5 presses, along with all press line I/O. The resulting distributed system consisted of over 500 parallel processes running on three networked QNX PCs, two of them embedded, with high performance event-driven inter-device coordination performed across the net.
The pilot successfully demonstrated a 26% increase in throughput, well above target, along with improved safety and ease of use. In addition, the high-level control language allowed our team to implement required control changes within minutes.
GENERAL MOTORS RESEARCH LABORATORIES, Warren, MI 1976 - 1992
Senior Staff Research Scientist (1988 - 1992)
Staff Research Scientist (1985 - 1988)
Throughout my career at GM, I made significant advancements and gained technological insights in engineering control. I built and led inter-divisional and inter-corporate engineering teams in developments that tested and demonstrated the value of theories that I developed.
In the course of control systems research, I invented a patented, self-learning technique, called adaptive feedforward, that would track a commanded path with extremely high accuracy, independent of commanded performance. When tested on robots, this control reduced peak errors by a factor of 1,000. The robotic community had believed the problem of speed-dependent error to be inherently unsolvable. The self-learning property eliminated variation, both from robot-to-robot, and from day-to-day for a given robot.
I discovered that command tracking control, which is highly stable, could apply to many control systems, such as vehicle control, previously envisioned as direct, rather than path-driven, control. I demonstrated that, for adaptive systems, real-time diagnostics are possible with no need for additional sensors.
I also discovered that high accuracy tracking control changes the emphasis from control design to real-time command path design. In response to this, I invented several algorithms for real-time computation of command path, based on solutions to minimum-time optimization problems, and demonstrated overall soft motion control (control loop computed entirely in software) on a number of systems. My patented techniques balanced algorithm complexity with real-time requirements. As a result of path planning, the tested systems showed increased performance and zero overshoot, along with high accuracy
In 1985, I was approached by the director of GM's Research Laboratories, to study vehicle design from a systems perspective. I co-authored an internal paper that characterized a systems approach to the development of a new vehicle, and produced a flowchart to describe that approach.
In 1988, I was granted two U.S. patents and received the "Roger B. Smith Extraordinary Contribution" award from GM, along with a generous financial bonus. I was the first mathematician at GM ever to receive a patent. Later, I acted as internal consultant on vehicle control within GM's research laboratories. I proposed a new direction in electronic vehicle control that would utilize an adaptive feedforward strategy and that could reduce variation, both from vehicle-to-vehicle, and from day-to-day for a given vehicle.
I also worked with engineers to design an adaptive feedforward approach for control of an electromagnetic engine valve, which works differently from a robot servo-motor, to eliminate valve clatter and associated wear.
I was invited to present seminars on adaptive feedforward and command path planning by the engineering schools of many universities, including The University of Southern California, The University of California at Santa Barbara, Stanford University, The University of Michigan, The University of Maryland, Harvard University, and Boston University.
Head of Mathematical Analysis Group (1982 - 1988)
Research Scientist (1976 - 1982)
I joined GM's Mathematical Analysis Group as a Research Scientist in 1976. Initially, I researched financial modeling of vehicle pricing under uncertainty and presented insights to the first annual General Motors Economic Marketing Conference.
In 1979, I joined a team engaged in the development of a hydraulic paint robot. I discovered that the robot had high following errors, and worked with the team to redesign system control algorithms. I invented a control that had both feedback (based on stochastic control theory) and feedforward components. We were able to reduce the error by a factor of 1,000.
In 1981, following extensive research, I built an inter-staff team of Manufacturing Development engineers to test a new control for a Unimation PUMA robot. As a result of our success, I made a proposal to Research Management and received a $1 million budget to perform advanced testing of the new robot control.
I then formed and led a larger inter-staff team to design and build controller hardware and software for testing on a GMF S-110 robot. We later broadened the team to include the Chevrolet / Pontiac group, with the intent to build a strategic alliance with an outside controller supplier, ultimately leading to commercialization.
I was Technical Leader in the entire partnering and development process, including bid evaluations, short-listing, interviews, contract negotiations, final selection of our two corporate partners - The Hughes Corporation and Modicon, Inc. - and hardware/software architecture design and coding. We received a direct GM budget allocation exceeding $3 million.
Our results with Hughes and Modicon controllers were similar to those of earlier tests. We reduced speed independent following errors by a factor of 1,000 (with peak errors under 0.005 inches); eliminated overshoot and improved cycle times; reduced spot welding test cycle time by 20% over the standard controller; and reduced path cutting test cycle time in excess of 60%. These tests showed that the new control would work on any type of servo-motor and that implementation could be accomplished economically.
PREVIOUS EXPERIENCE
WAYNE STATE UNIVERSITY, Detroit, MI
Lecturer - Computer Science Department (2003 - 2004)
Taught classes in C++, Discrete Mathematics, and the Structure of Programming Languages.
OAKLAND UNIVERSITY, Auburn Hills, MI
Adjunct Faculty - School of Engineering (1985)
Invited to teach a graduate course in Large Scale Systems Theory.
UNIVERSITY OF DELAWARE, Newark, DE
Visiting Assistant Professor - Mathematics Department (1975 - 1976)
Taught classes in Operations Research, Probability Theory, and Advanced Calculus.
PATENTS
"Coordination of Computer-Controlled Events Involving Repetitive Use of Servo Devices and Presses," U.S. Patent Pending.
"Integrated Motion Control Algorithm with High Position and Temporal Accuracy," U.S. Patent Pending.
"Compiler Design Using Object Technology with Cross Platform Capability," U.S. Patent No. 6,305,009, October 16, 2001.
"Motion Control System with Minimum Time Path Generation," U.S. Patent No. 4,769,583, September 6, 1988.
"Motion Control System Having Adaptive Feedforward Path Tracking," U.S. Patent No. 4,761,595, August 2, 1988.
SELECTED PUBLICATIONS
"New Coordination Algorithms Improve the Throughput of a GM Press Line," Proceedings of the International Robots & Vision Conference, Chicago, IL, June 2003.
"Controlling the Line," QNX News, Vol. 11, No. 2, pp. 15-19, 1997.
"An Adaptive Feedforward Approach to Robot Control," Mathematics in Industrial Problems, Part 4, Avner Friedman, ed., Springer-Verlag, New York, NY, Chapter 20, pp. 186-193, 1991.
"A New Approach to Minimum Time Robot Control," Robotics and Manufacturing Automation, M. Donath and M. Leu, eds., American Society of Mechanical Engineers, PED-Vol. 15, pp. 1-11, 1985.
"A New Approach to Robot Control," Proceedings of the American Control Conference, pp. 387-389, June 1985.
TECHNICAL SKILLS
Proficient in: Mathematica, Microsoft Office, Microsoft Visual C++ (VC++), Gnu compilers, UNIX, Microsoft Foundation Classes (MFC), and simulation.
Mathematical expertise in: Ordinary differential equations (ODE), partial differential equations (PDE), stochastic (Ito) calculus, numerical methods, and math models.
EDUCATION
UNIVERSITY OF MICHIGAN, Ann Arbor, MI
Ph.D., Mathematics
M.A., Mathematics
HARVARD UNIVERSITY, Cambridge, MA
B.A., Mathematics |
