Serpentine Robot Arm with Tendon Transmission
With 8 main joints, continuously rotating base, and wine glass gripping hand.  Mainly 2001-2003.  Max Height = 96". Actual arm length above platter = 48".

Machined Aluminum, polycarbonate, steel; ball bearings, cable, DC gearmotors, custom electronics, sensors, lighting, software, etc.  Design, fabrication, firmware, software, and electronics - Carl Pisaturo.
An  exploration of the aesthetic and technical realities of a multijoint, tendon-driven life-like structure

A  development laboratory for  techniques to be used in later kinetic works
The ostensible function of the machine is the artful handling of wine glasses.  This includes locating, grasping, nimble transport, and hand-off of a glass of wine to a person, without spilling.  The wrist, "hand" and several sensors are specifically designed for this mission,  and a sonar sensor is used to locate people who need a drink. 
The machine is fully motion controlled - that is, joint positions and velocities are known and controlled by system software.  Any movement, within the limits of the structure, can be designed and executed later, or dictated "live" by some higher software process.  Problems such as collisions and mechanical malfunctions can be sensed and dealt with on the fly. Accelerometers in the wrist tilt the wine glass correctly into any motion to prevent spilling. There are 30 sensors, 12 actuators, 2 lighting circuits and a speaker on the arm.
The Serpentine Arm is a cable-tendoned stack of 8 hinge joints which rides on a continuous rotating base, and that rides on a fixed base. The tendons revector (pass through joint jogs) on ball bearing pulleys.  At the arm's end is a roll / pitch wrist and a special wine glass gripping hand.  The joint arrangement of the arm itself is planar, like 8 links of a bicycle chain. Each of these links is individually controllable through plus/minus 60 degrees. The arm can move quickly, and can assume a wide range of postures. The arm itself (not counting the motor deck and pedastal) is about 4 feet long and can lift about 1 pound.
Perhaps the biggest reason for this project was to learn, evaluate and evolve some technologies that will be useful in later kinetic sculpture projects.  Among these:
* Remote placement of heavy motors off-limb to achieve smaller/faster/nicer limb, with cable transmission of power from motor to joint, over ball bearing revector pulleys.
* Slip-Rings (passing of power and signal lines through a continuously rotating joint).
* Serial HC12 network design, to achieve modular distributed processing.
* Implementation of true servomotor control with microcontrollers.
* Use of Windows / Visual Basic programs on laptop as robust, top-level software.
* Programmable lighting.
The animal kingdom humbles as much as it inspires the robotic engineer. Animals set the standard of perfection, and taunt the tool-using human to just try it. Those who do soon confront the unforgiving and unbullshitable laws of reality. And even when a long series of battles is won, human creations of the life-like are crude, dumb, and delicate.   Still, there is a strange beauty when metal and plastic structures try to live.  Human pride wants better robots.
This project proved to be quite complex, especially on the firmware and software side - maybe too big a project for one unfunded person. It was actually several non-trivial projects interacting with each other in complicated ways.  Although most of the main milestones have been achieved, there is no clear "done" point.  Along the way, major improvements become obvious, often expensive in time and money to implement. Clearly, the evolution of this system could absorb any amount of time and money.  The ultimate goals, as usual, are animal-like grace, high reliability, easy rich programmability and robust protection from itself and environmental collisions - a tall order indeed.  Most likely, the project will reach a state of "robust, reasonable showability" and stop there.
Control Sys
Serpentine Arm - Overview
works of  Carl C Pisaturo