The design of machines that rely on arrays of servomotors such as robotic arms, orbital platforms, and combinations of both, imposes a heavy computational burden to coordinate their actions to perform coherent tasks. For example, the robotic equivalent of a person tracing a straight line in space requires enormously complex kinematics calculations, and complexity increases with the number of servo nodes. The conventional method of executing these calculations is with a PC-style set of electronics including a powerful CPU (central processing unit) microprocessor, operating system, power supply, a number of peripherals, connectors to support each servo node, and a web of startopology wiring across the machine (including flexing joints), generally exceeding 100 conductors. In industry, the most common implementation is one or more dedicated PC cards mounted on an ISA (Industry Standard Architecture), PCI (Peripheral Component Interconnect), or VME bus. These cards provide the I/O connectors and supplement the CPU to execute the massive kinematic calculations in real time.
A new high-level architecture for coordinated servo-machine control enables a practical, distributed transputer alternative to conventional central processor electronics. The solution is inherently scalable, dramatically reduces bulkiness and number of conductor runs throughout the machine, requires only a fraction of the power, and is designed for cooling in a vacuum.
The benefit of this innovation is total elimination of the central controller and reducing the heavy web of star-topology wiring across the machine to four wires along a shared serial bus. This scalable innovation results in decreased power consumption, decreased bulk, and vacuum compatibility.
This controller places a digital signal processor (DSP), instead of a microprocessor, at each motor axis, each with its own power supply, conduction cooling, etc. The DSPs communicate via CANbus over RS-485 hardware, forming the heart of the transputer.
Features of the device include reduction in flex wear in serial-articulating joints, streamlining of robotic structures (with reduced wire-harness bulk), survivability in the case of a single-processor failure, and uniformity in handling various types of servos (brushless, brushed, etc.) and sizes up to 300 W. Brushless compatibility supports elimination of brush-life limits and particulate generation. Power in the 2-wire bus flows directly between regenerative power nodes and motive power nodes, rather than traversing the round-trip star topology.
This work was done by Brian Zenowich, Adam Crowell, and William T. Townsend of Barrett Technology Inc. for Johnson Space Center.
In accordance with Public Law 96-517, the contractor has elected to retain title to this invention. Inquiries concerning rights for its commercial use should be addressed to:
Barrett Technology Inc.
625 Mount Auburn Street
Cambridge, MA 02138-4555 USA
Fax: (617) 252-9021