Home Engineering Audio Biking Publications Photo Gallery Personal Info
Engineering Projects: Robotic Arm (Phase IIIa)
  [Phase III arm]

Photo of the Phase IIIa/IIIb arm, under development. The Phase II arm and table were constructed well enough that they served as the platform for continued development. Phase III proved to be the most productive phase of development. In this phase, a complete working image acquisition and recognition system was created from scratch. The control electronics were improved, the software rewritten, and impressive automated arm applications were developed. The latter third of Phase III saw the complete replacement of the wood arm with an aluminum one. The Phase III system was arguably the most impressive as it showcased a very tightly integrated vision, decision-making, and control system.

Vital Statistics

Years Summer 1991 - February 1992
Construction Materials Wood, carpenter's glue, sheetrock screws, nylon mason's line, threaded rod, bearings, surplus satellite dish rotator motors, misc. surplus motors, misc. plumbing parts
Degrees of Freedom Four plus grip (linear base, shoulder, elbow, wrist). Cartesian (box) work envelope.
Drive Techniques Winching nylon lines around a pulleys on DC gear motors to raise joints. Gravity needed to lower joints. Base used a feed-through style winch with tensioning turnbuckle. Wrist pitch motion accomplished via chain drive with 1:1 drive.
Feedback Three potentiometers (shoulder, elbow, wrist). Optical interrupter sensor for linear encoder (base). Limit switches on base, gripper. Rudimentary analog pressure sensors on gripper pads using conductive IC foam.
Control Computer 12MHz 286 PC with VGA graphics and 120 megabyte hard disk.
Interface An external interface cabinet with an internal card cage contained the power supplies, motor control, scanner interface, etc. A front panel display with LEDs and a 16-character LCD provided status information to the operator. A new PC ISA card with 72 digital I/O lines connected the PC to the external interface cabinet.
Software Written entirely in Borland Turbo Pascal 6.0 using BGI graphics for custom GUI.
Image Acquisition Roughly 40 inch wide by 16 inch high scan area, using an opaque object detection scheme where shadows of objects placed on scanner bed were detected. Resolution of approximately 1/4" by 1/2" accomplished with linear encoder and spacing of scanner head elements (Cadmium Sulfide photoresistive cells). Single bit scan depth with hardware comparator setpoint for each scanner head element used for PC interface.
Capabilities Automated object retrieval using image scanner system with image processing and intelligent control application. This was the high point in the development of the entire system, as it reliably demonstrated interaction between the image acquisition and arm motion control systems. Automatic pick and place operations as well as joystick-driven manual operation were also implemented.


[Phase IIIa video footage] [Low Quality MPEG1 320x240, 5.8MB]

Additional Photos

[Phase III table renovations] [More phase III table renovations] [Phase III interface cabinet in construction]
This photo shows the newly-renovated table before painting and plexiglas. The table from Phase II was used with significant modifications for the image scanner. The right half of the 1/4" plywood surface was entirely replaced and a new under-table portion was added to protect the scanner. Note the two bearings captured in the wooden guide rail on the rear. The two protruding rods on the base of the arm mate to the bearings to provide smooth linear motion. A bit further along in the Phase III portion of the project, you can see the nearly completed new scanner area. The vertical supports at each end of the table supported overhead lights for scanning (initially two incandescent, eventually a fluorescent). In the large version of the photo, the linear encoder is visible on the left. The final coat of paint had been applied to the new portions of the table and the plexiglas is also installed. You can also see the arm from Phase II (sans gripper) being prepped for Phase IIIa work. The interface cabinet and card cage after assembly and before painting. Note the 1/4" plywood inlay on the faces to facilitate mounting of switches, controls, etc.
[Phase III interface cabinet] [Phase III interface cabinet] [Phase III interface cabinet]
The interface cabinet, much later in development, showing several cards in the card cage. Also clearly visible is the Black & Decker power supply used to run the motors in the arm. The rear of the card cage. Clearly visible is the potentiometer bank, used for adjusting the thresholds of the comparators for the image scanner. Also shown at the top are the "ice cube" style relays used to control the motors in the arm. The front of the interface cabinet, showing the power switch, key switch, emergency stop button, LED status panel and 16x1 LCD.
[Phase III interface cabinet] [Phase III interface cabinet] [Phase III image scanner]
The power supply section of the interface cabinet. A four-output custom supply was used to drive the logic and relays while a surplus Black & Decker adjustable DC supply from a Dremel-like tool was used for motor drive. The potentiometer bank, comparator circuits, and multiplexers used to run the 32-element linear array image scanner. The 32-element linar array scanner. Small 1/4" PC board mount Cadmium Sulfide (CdS) cells were used, arrayed on two hand-made PC boards. Vertical resolution was approximately 1/2". The scanner functioned similarly to a photocopier or flatbed scanner, except instead of using reflected light, it detected the shadows cast by objects placed on the plexiglas surface. The hand-made linear encoder is also shown in this photo. It was made out of 1" X 1/16" aluminum angle stock and provided roughly 1/4" resolution. Drilling out two of these (the other was used for the linear base motion of the arm itself) was one of the least exciting parts of the project.
[Phase IIIa arm modifications] [Phase IIIa overall development shot] [Phase III scanner element closeup]
This shot details the chain drive for the wrist roll motion when it was under construction. The linear drive motor and barrier strips for wiring are also visible in the shot. The full arm system, mid-development. Note the alternate wrist design which was under test. The interface cabinet, 286 PC and monitor are also visible. Closeup of the scanner element, showing the 1/2" spacing of the CdS cells.
[Phase III/IV ISA Card]    
ISA interface card used in Phase III and later in Phase IV. The card provided 72 digital I/O lines which connected to the external interface box. The card underwent some minor changes at one point several years after Phase IV for the SMSurround project. You'll notice slight differences between the picture above and the the picture in the software screenshot below.

Software Screenshots

[RACS 3.x GUI] [RACS 3.x GUI]
The Integ integrated control system application, complete with custom GUI. The software was written in Turbo Pascal using the BGI graphics primitives library. The GUI bore the contrived acronym SIGMA (Software Integrated Graphical Mouse Application). Note the "automated pick-up" option on the right screen shot. This launched the fully automated retrieval operation. The scan data shown in both screenshots is bogus. Unlike later versions, this software did not allow scans to be saved and loaded, so no valid scan data could be displayed for these screen captures.
Table of Contents
Robotics Home
Phase I 1989-1990
Phase II 1990-1991
Phase IIIa 1991-1992
Phase IIIc 1992
Phase IV 1992-1993
Page last modified:
Copyright © 1993-2003 prefect - All Rights Reserved.