STIC Insect by Univ of S. Queensland, Australia

Organisations:
University of Southern Queensland, Australia
National Centre for Engineering in Agriculture, Australia

Project Supervisor:  Professor John Billingsley, PhD, MA (Cambridge)
2nd Supervisor:      Dr. David Hilton, PhD, BE
Project Engineer:    Sam Cubero Jr, BE Hons (Mech), PhD candidate
Workshop Technicians: Mr. Peter Penfold, Chris Galligan
Purchasing officers:  Lez Seal, Adrian Blokland
Administration:       Malcolm McKay, Prof. Peter Swannell, Mr. Ron Ayers


Project:

SPACE TRUSS INTEGRATED-CONSTRUCTION ROBOT:  "STIC Insect"

(Patents pending worldwide - Floating Plate Gas Valve & Mechanical 
structure)


For general information and photos, see URL page:

http://neptune.eng.usq.edu.au/~cubero


 Complete description:


      Functions:

The STIC Insect walking robot is a pneumatically powered, four legged
vehicle designed for walking over unstructured surfaces, travelling
from a flat floor onto a flat vertical wall, and travelling from an
outside flat vertical wall to a higher, outer horizontal rooftop.
(internal and external corner transitions).  It has the ability to
rotate on the spot to change its heading, move forwards and backwards
and even sidestep.  It is not yet complete, but is expected to be
walking by the end of 1996.  (Two legs are working fine at the moment)
The robot may also be powered using hydraulic cylinders, linear DC
motor drives (ballscrew type), and other translational type drives
since the connection points between adjacent links are rotary pivots.
The use of linear DC motor drives would provide the most accuracy,
and hydraulic cylinders would provide similar levels of accuracy but
with greater power.  However, pneumatic actuators were selected due
to the requirements for a very lightweight structure and simple compliance
control for each of the robot's legs.  DC motor driven ball screws with a
high gear ratio (small thread pitch) are not easily backdriven, and fluids
cannot be compressed unless bulky and heavy gas filled accumulators
are used.


      Specifications:

The STIC Insect walking robot is described in 3 technical papers:

1.  "Automatic Surface Transition Adaptation for a Quadrupedal Space
    Frame Robot" -Proceedings of the Mechatronics and Machine Vision
    in Practice II Conference, Hong Kong 1995, ISBN 962-442-076-9

2.  "A Novel Proportional Gas Valve for Mechatronics Applications"
    -Proceedings of the Mechatronics and Machine Vision in Practice II
    Conference, Hong Kong 1995, ISBN 962-442-076-9

3.  "Automatic Control of a Surface Adapting, Four-legged, Wall
    climbing Robot" -Proceedings of the Mechatronics '96 Conference,
    5th UK Mechatronics Forum International Conference with the 3rd
    International Conference on Mechatronics and Machine Vision in
    Practice, Volume 1 page 135, Portugal 1996, ISBN 972-8063-08-3

The mechanical structure of the STIC Insect is comprised entirely of
tetrahedral pyramids, providing high rigidity with minimum weight.
Double acting pneumatic air cylinders are used to provide variable
force and position control for each link, hence, each foot can be
assigned a unique compliance ("springiness") vector for any position in
its three dimensional workspace.  This is important during passive
foot searches which involve collisions on walls or undefined surfaces.
Another area worthy of investigation is autonomous navigation using
vision guidance.


      Parts used:

Mechanical hardware:
16x "SMC" double acting pneumatic cylinders (varying strokes)
96x hollow alluminium tubes, 12mm dia., 1.2mm wall thickness
192x threaded aluminium end connectors
32x 8mm shaft deep groove ball bearings
100+ custom fabricated and welded aluminium components (50+ drawings)
16x springs for the passively compliant, self-centring robot feet
48x "Floating Plate" proportional gas valves (patents pending) for
   pressure and flow control.  (custom designed for 0-25 LPM flow)
1x 13CFM air compressor
1x 0-100LPM air flowmeter
4x on-off solenoid valves for the vacuum feet
200+ air fittings, tubes, nuts & bolts, screws etc.

Feedback sensors:
12x Single-turn rotary potentiometers for position sensing
24x "Sensym" 0-100psi Pressure sensors
4x  Vacuum pressure sensors
8x  Momentary contact foot switches

Control hardware:
1x 486 PC computer with COM1 serial port
8x 8-bit single-chip controllers with various accessories and circuits
8x RS232 communications interface chips connected to the COM1 port
1x dual DC power supply (power: +15V @6A , +5V, GND)

Software:
1x Master control program / simulator with Graphical User Interface
   (about 250 Kbytes in .EXE form, or 150  A4  pages with comments)
12x 8-bit assembly language programs for low level axis control
   (each program being about 60  A4  pages long with comments)


      Problems encountered:

Too many to remember but they've all been solved, thankfully!
Air is extremely compressible and "weak", so much effort has been
spent in developing and refining new "Floating Plate" proportional
gas valves for achieving crisp, rapid and accurate position control.
Commercial on-off air valves were too noisy, slow and wasteful of air.
We have developed a "smart cylinder" which minimises air wastage
during force or position control of a piston.  We are seeking manufacturers
and investors who would be interested in commercial exploitation of this
new type of actuator as it has the potential to be the cheapest form of
programmable axis on the market which offers force and position control.
A non-linear, adaptive gain, PD-type control algorithm is used for
controlling the robot's cylinders to minimise oscillations which may
occur, depending on the geometry of each leg.  The inverse kinematics
of the robot was probably the most difficult part of the project to 
complete,
however, a robust numerical solution involving the inverse Jacobian was
formulated and proven to provide satisfactory, real-time foot control in
world space coordinates.  (see papers 1 and 3)


      Time to build:

2 years and 9 months to date, with just myself as the designer and
researcher.  (5 months of University workshop labour was required for
the production of the majority of mechanical components)


      Cost:

Parts:  Approximately A$10,000
Workshop labour:  A$12,000
PhD scholarship:  3x A$18,500 p.a.
Total:  Approximately A$77,500.


      Other important information:

The robot should be up and walking about in less than 2 months time.
The main motivation behind this work was to develop articulated link
manipulators which can be controlled accurately in force and position
so that they may be implemented as high speed robotic fruit pickers.
Several applications exist in Australian agriculture which require
robots to pick fruit, vegetables and tabacco leaves in a very
delicate yet high speed manner.  Rockmelon harvesting is one possible
application for the legs of the STIC Insect robot.  A 14 metre wide
(tractor like) gantry has been built at the University of Southern
Queensland for the purpose of harvesting rockmelons using machine
vision techniques.  It will be used as a testbed for future
experimental work on automated farming and agricultural operations.
If you wish to picture it, imagine a row of 14 STIC Insect legs in a
line, testing the ripeness of each fruit using in-line gas sensing.
Each manipulator carries a tool which can twist a ripe rockmelon
off its vine for placement onto a conveyor belt, which transports
the fruit to a large onboard container.  The 14 fruit pickers are
supervised by an operator who ensures that the machine vision steering
system keeps the gantry vehicle on track with respect to the furrows,
within which the hydraulically powered wheels move.  I'll provide a
picture of this monster robot when it is built and operational.

With the "machine vision" extertise of Professor John Billingsley,
(designer of the "Steeroid" automatic tractor guidance systems at the
NCEA - National Centre for Engineering in Agriculture, Australia)
this robotic testbed is no longer a dream away.  This technology will
affect the food and agricultural industries in a very big way,
perhaps even as profoundly as it had affected manufacturing.  It is
my hope to see this dream come true.  I encourage all young and old
to get involved with mechatronics and robotics engineering.  This is
where the real world "action" is happening and it is definitely the
way to a more prosperous and exciting future for everyone involved.

Keep your dreams alive!