Robotics
The Robotic revolution
Robotics is one of the most rapidly-expanding fields in Engineering today. Once thought of as machines to replace repetitive labour, development at research institutions such as MIT is bringing the concept of the 'intelligent' machine closer to reality. This is most useful in the areas where human access is impossible, or the design concept is so complex that some element of learning must be implemented.
This page is divided into the following sections:
The Project -
Information regarding the initiation of the project
Specifications -
Physical and performance characteristics of the robot
Construction -
Materials used in the process of construction
Description - A
brief yet concise description of various aspects of the robot
Why a Hexapod? -
Motivation for creating this beast
Links and References - Links to other sites, and references for material used
The Project
In the year 1997, in the 7 month period from March - September, a
friend and I went about constructing a six-legged walking machine, known as the
Robobug. It was initiated by Dr Grant A. Covic at the University of Auckland's
Electrical Engineering Department as a fourth-year project. The project was given a
$200NZ budget ceiling, making design decisions critical. This is not a lot of money
to work with, when compared to other projects.
Specifications
Leg:
Weight
: 300g
Overall dimensions
(WxHxL) : 80 x 280 x 175mm
Maximum lift
height
:110mm
Maximum lift
depth
: 145mm
Maximum turn
angle
: 160 degrees
Robot:
Dimensions
(HxLxW)
: 280 x 280 x 370mm
Theoretical top speed using
Tripod
: 0.308 m/sec
Phase
: 0.123 m/sec
Wave
: 0.062m/sec
Practicable top speed using
Tripod
: 0.10 m/sec
Phase
: 0.04 m/sec
Wave
: 0.02 m/sec
Operating
voltage
: 4.8 - 6V
Power consumption
(idle)
: ~2.5A
Power consumption (phase mode) : ~5A
Stationary height above ground :
110 mm
Maximum height above ground :
145 mm
Total
Weight
: 2665 grams
Servo Motor Specifications:
Torque 4.8V
(kg.cm)
: 3.1
Torque 6V
(kg.cm)
: 3.5
Speed
(msec/10)
: 4.17
Sweep Spd
(msec/10)
: 3.67
Weight
(g)
: 41.67
Size WxLxH
(mm)
: 19 x 39 x 33
Idle
Current
: 4 mA
Motor
: 3-pole Ferrite
Operating
voltage
: 4.8 6V
Average current
draw
: 0.4A
Maximum current
sink
: 1A
Construction
~1.5m of 9mm x 9mm aluminium bars
12x JR-NES 507 servomotors
280mm x 140mm perspex base
130 x 130 main PCB, with Philips 80C552 microprocessor
6x Leg controllers, with Atmel 89C2051 microprocessors
Fibreglass shell
Perspex pieces, rubber and switches for the feet
Assorted hinge joints, screws, and washers
Countless hours or hard labour, sleepless nights, and cups of coffee
:-)
Description
The Robobug is a hexapedal, mobile, robotic insect consisting of a
modular assemblage. It is fully operational and was constructed in the seven month
time frame to all specifications. It currently has the ability to walk forwards,
backwards, and turn left or right. Control was achieved from a computer terminal via
a serial link. An graphical-user-interface was designed using Borland Delphi for
this purpose. The performance of the system was good in terms of resources consumed
and stability of operation when compared to other efforts. For
starters, it can lift it's own weight on power up (an achievement when you look at other
efforts!) and was completed by two undergraduate students. Not bad when you
consider that other robots were worked on by teams of up to 20, in a much greater
time frame, with a much larger budget. This type of project is also normally
reserved for postgraduate students. Apparently we were the first group to
successfully and complete a prototype robot in seven months, at the University.
Doesn't say much for the uni I guess! :) (that was a joke Dr Covic!)
There are three different modes of walk; Tripod, Phase and
Wave. These employ 3, 4, or 5 legs on the ground, respectively at any one
time. They give different degrees of stability and speed. Perhaps the best
part of the robot is its ability to turn. It takes around 4 seconds to turn 90
degrees, which is relatively fast for a legged robot.
Some of the outstanding features of the design include:
1. The mechanical leg's large work-envelope and unrestricted placement. A typical four-bar mechanism was used for good stability. A pantograph mechanism could have been used to allow linear motion, although the budget of the project prevented this.
2. The utilisation of independent leg controllers distributed the computational load on the central microprocessor. Unused resources were thus made available to the more CPU-intensive tasks of the future. The control software was written entirely in 8051-based assembly language code.
3. Movement was directed using a vector-based behavioural control method. This simplified obstacle avoidance by using vectors to model the robot's course and immediate surroundings.
4. The Robobug provides an excellent platform for the development of feedback and control mechanisms to promote intelligent behaviour. These will allow the robot to acquire perception by continuously monitoring its environment and making logical decisions. This attribute is essential in order to achieve true autonomous navigation.
There are several improvements which could be made. For a start, the robot can take very little addional load. This means sensory equipment, used to inform the robot about its surroundings, difficult to add on. This is easily fixed by purchasing servos with a higher torque rating. In addition, the leg design places too much torque on the lift-servos. Additional development efforts could include sonar or infra-red sensors for obstacle detection, bump sensors on the sides, and a more sophisticated controller for a neural net or similar learning control method. Although ground contact sensors were installed on the bottom of each foot of the robot, these were not implemented due to time constraints. They will be used to walk on sloped or irregular ground.
Unfortunately I don't have a scanned picture of the robot to show you at the moment. I will post one or two as soon as I get them.
Why a Hexapod?
You may be asking yourself : Why build a legged robot? What's wrong with wheels? Sure, conventional wheeled robots can move much faster than wheeled robots, have greater weight-carrying capacities, and are a lot easier to build. But legged robots have the advantage in that they can be accurately controlled to walk over uneven terrain, move up stairs, or over obstacles. A wheeled robot, such as the Mars Rover, is restricted in that it can only move over gaps in the ground half as large as its wheel. This makes legged 'bots better for moving over cracks, like into the heat of a volcano, around rocky planets, or in hazardous waste disposal situations. The construction and simulation of a Robobug also allows us to model the movement of insect motion, and possibly behaviour as artificial intelligence is added. For example, a group of Robobugs could work collaboratively to detect and disarm landmines in an area, quickly and safely.
Links and References
[1] Mobile Robot Research at the MIT AI Laboratory, MIT Robotics
Laboratory, Cambridge, MA, 1997
http://www.ai.mit.edu/projects/mobile-robots/
[2] Scott, F., Frank Scott's Hexapod Robots, Sutton, London, 1997
http://www.frasco.demon.co.uk/#Intro
[3] Horizon Hobby Distributors, Inc., JR Servo Specifications,
1997
http://www.horizonhobby.com/jr_folder/jr_sevro_specs.html
[4] AT89C2051 Microcontroller Data Sheet, Atmel Corporation, 1996
http://www.atmel.com
[5] Balkenius, C., Kopp, L., & Pallbo, R.,
LUCS Autonomous Robots, Lund University Cognitive Science, Kungshuset/Lundagerd, Sweden,
1995
http://lucs.fil.lu.se/Projects/Robot.Projects/robot.proj.html
[6] Berns, K., The Walking Machine Catalogue,
1997
http://www.fzi.de/divisions/ipt/WMC/walking_machines_katalog/node12.html
*The walking machine catalogue [6] is an awesome site to go to for finding specs on other legged robots.
Return to Main page