Due to the general decline of manufacturing industries in America, and also the knowledge and cooperation made possible by the internet, computer-controlled machine tools are undergoing a quiet but profound shift in the way they are used. The process of transformation from giant industrial machines to small and convenient personal tools is very similar to the beginning of the personal computer movement. The personal computer was made possible by the availability of cheap silicon micro-circuits, and it will likely require a similar profound change in the way machine tools are built in order to reach the much talked about "fabber in every basement" utopia. Currently in America we are cutting down the last of our forests and disassembling our factories to be sent to China and turned into cheap disposable products, and then they are shipped all the way back across the Pacific Ocean to be consumed and ultimately discarded. However, most of these products can be made locally, economically, to high-quality, conveniently, from durable recycled materials, and to personalized specifications or from standardized plan files downloaded for free from the internet, all done by miniature automated factories kept in our own homes. This proposal is a step on the path toward that goal. The variable-strut-length octahedral parallel kinematic manipulator, or in short a "hexapod", is to conventional machine tools such as milling machines as geodesic domes are to conventional buildings. (milling-machine.jpg) The hexapod architecture, or "stewart platform" as it is known outside of a machine tool context, is currently used in many diverse fields for: alignment of mirrors and laser optics, nano-scale positioners, driving and flight simulators, landing gear testing, and orthopedic braces. The hexapod substitutes a triangulated frame for the square stacked blocks that make up the structure of an ordinary machine. This triangulation provides extremely high rigidity for a given mass of steel, making a very convenient and portable tool with the same or higher capabilities as a conventional machine. In general, they are lighter, faster, stronger, and more accurate than a conventional tool. Furthermore, the hexapod is much easier to make and distribute because most of the complexity and precision is taken out of the manufacturing process and moved into the realm of software and electronics. Traditional machine tools are expensive to make and distribute because they: 1) contain large heavy pieces of metal - complicated shipping; more inertia requires bigger motors 2) have many precision surfaces that must be in near-perfect alignment - difficult to manufacture 3) are usually hand-finished to achieve the necessary level of precision - skilled labor is expensive 4) are made of many different complicated shapes - more design effort, hard to pack for shipment 5) distort unpredictably due to non-kinematic design - compensated for with expensive materials 6) defects in one axis change the position of axes on top of it - all axes must be more accurate 7) do not make full use of material strengths and motor power output - most of the machine sits idle 8) must be placed on a sturdy foundation and perfectly leveled By contrast, a hexapod can be made inexpensively because it: 1) is relatively lightweight; disassembles into easily packed shapes - prototype fits in a box (box.jpg) 2) can correct almost all types of manufacturing errors in software, so alignment is a non-issue 3) can be calibrated automatically with simple equipment [ishida] 4) contains many identical small parts, and only one type of precision part - the screw 5) is constrained only by point contacts in the mechanism; little bending in the structure 6) errors are averaged out - total positioning error is roughly one half of the error in an each strut 7) utilizes all six motors at once, and all structural members share loads equally 8) can be set up anywhere, even outdoors or hanging from a tree! Why low cost machine tools? Trying to promote advanced machine tool technology to the general public is frustrating. Most people will concede that, although manufacturing technology advances have some intangible benefit for "our" industry as a whole, they have little direct effect on daily life. Advances in machine tool technology only contribute to the economic power of the small segment of the population that can actually afford new machine tools. However, if they can be made cheap enough, we can simply distribute machine tools to whoever wants them. The only requirement is that the machines be inexpensive enough for the average person to buy outright. The very few hexapods currently in existence are either research machines built from scratch by well-funded university labs, or experimental production tools that cost millions of dollars. In order to be relevant to the largest segment of humanity, and to improve the capability of individuals everywhere, my first priority is to develop a machine tool that is affordable to the common person. When the solid struts are replaced with steel cables, the device is known as a "RoboCrane" [albus] and larger devices on the scale of tens of meters are possible. These six-axis cranes would be much more positionable than a regular crane, and stiffer in all directions and rotations. They could be used for: structural welding, sandblasting, automated construction including "contour-crafting" [khoshnevis], skateboard park construction, aircraft and building maintenance, virtual reality haptic interfaces, loading fallen timber, pulling fallen buildings off of people, automated greenhouse tending, fruit and vegetable harvesting, film and television camera positioning, physical therapy tools for stroke victims, and many more uses I don't have space to list. Progress towards cheap hexapods so far: The Laboratory for Micro Enterprise, a web site by John Storrs, made a significant contribution to the low-cost hexapod concept in the late 1990's with detailed plans and simulation code for a machine designed for small woodworking businesses. [storrs] Carl Mikkelsen constructed a light duty hexapod in his basement and was able to reliably machine aluminum and wood. [mikkelsen] Some of the software that was developed in the original NIST/Ingersoll research project [nist-report] (nist-hexapod.jpg) was released into the public domain. It has since been greatly improved by volunteers from around the world, and is now part of a thriving real-time Linux software project for machine tools of all types. [emc2] I have recently added a simulation/visualization component to emc2 for improved cooperation and bug checking between physically distant software developers. (hexagui.jpg) I have also designed a complete inexpensive servo motor control system from commodity parts (motherchip.jpg) and designed a motion platform, easily assembled octahedral frame, enclosure, and actuators. (hextatic.jpg) So that it can be distributed widely for no charge, easily improved by anyone and adapted for use in unanticipated contexts, all of my work has been done with free and open source software tools. Ongoing development progress is documented on my DIY machine-tool wiki. [hextatic] I estimate my design for a complete turn-key desktop system to cost around $500 USD shipped to your door in two small boxes. Currently, computer controlled desktop milling machines of dubious quality will cost a minimum of $2500 if bought outright, and $40,000 is an average price for a low-end full sized mill. Work that still needs to be done: Build and measure performance of the prototype hexapod kit. Testing of servo interface and power electronics. More testing with kinematics transformation calculations in real-time on slow computer hardware. User-friendly software configuration with emc2, description of a well defined work envelope so as not to crash into the frame, and more work on the emc2 motion planner internals. Start a business and investigate legal issues. Research into viscoelastic shear damping, resonance detection, and automatic calibration - any of which may drastically improve performance at minimal cost per device. Development is an ongoing process and it is unlikely we will ever see a "perfect" hexapod anywhere. What's a trimtab got to do with it? I believe that if we can get a relatively small number of these machines into the hands of people that use them daily, then the experience of seeing them being successfully used will ease much of the uncertainty and doubt about hexapods that is prevalent in the manufacturing community. This process can be accelerated through internet self-publishing by users of the machines. Once hexapods are accepted as a viable alternative, they will become abundant because of their convenience and reduced cost, with many independent manufacturers each producing their own design. I propose to create a small business selling hexapod "kits" in order to begin this process. Unfortunately, the typical route of venture investment has a high likelihood of corrupting the project's spirit and purpose, which will ultimately limit its impact. Without a benefactor, I will be not be able to concentrate fully on the development and promotion of these tools. ---- References ishida: http://mmc.me.kyoto-u.ac.jp/research/para/ishida/pa_ishida.html albus: http://www.isd.mel.nist.gov/projects/robocrane/ khoshnevis: http://www.isi.edu/CRAFT/CC/modem.html storrs: http://www.laboratoryformicroenterprise.org/lme/LMEHexapodMachine.html mikkelsen: http://www.foxkid.net/cmm/platform/project-notebook.html nist-report: http://www.isd.mel.nist.gov/documents/wavering/PKM_Final.pdf nist-hexapod: http://www.mel.nist.gov/photos/hexa.html emc2: http://www.linuxcnc.org/ hextatic: http://fennetic.net/machines/?hextatic