The University of Western Australia
Loading the player ...
- Offer Profile
|Welcome to the Robotics and Automation Laboratory at the University of
Western Australia, Perth. The Lab has been active for over a decade doing
research on all types of autonomous mobile robots, including intelligent driving
and walking robots, autonomous underwater vehicles, and unmanned aerial
vehicles. We also work on the design of embedded controllers and embedded
operating systems, as well as on simulation systems.
Embedded controller for and robotics applications, digital camera, sensors,
- EyeBot is a controller for mobile robots with wheels,
walking robots or flying robots. It consists of a powerful 32-Bit
microcontroller board with a graphics display and a digital grayscale or
color camera. The camera is directly connected to the robot board (no frame
grabber). This allows to write powerful robot control programs without a big
and heavy computer system and without having to sacrifice vision - the most
- Ideal basis for programming of real time image processing
- Integrated digital color camera
- Large graphics display (LCD)
- Can be extended with own mechanics and sensors to full mobile robot
- Programmed from IBM-PC or Unix workstation,
- Programs are downloaded via serial line (RS-232) into RAM or
- Programming in C or assembly language
- Third generation hardware
Autonomous driving robots, differential drive, EyeBot controller, digital
- Eve (EyeBot Vehicle) was the first driving robot we build
around a specialized EyeBot controller and a QuickCam camera system. This
robot has the standard top part of the EyeBot M1 controller, but a modified
bottom board that matches the outline of the physical robot. We later
discarded this technique for a standard controller (M3) that is identical
for all robot vehicles.
Eve is equipped with:
2 DC motors with encapsulated gears and encoders
1 Infra-red PSD sensor
6 Infra-red proximity sensors
Acoustic bumper system
QuickCam digital camera
- Our next driving robot designs were a number of
variations for the CIIPS Glory Robot Soccer Team. These robots had to be
somewhat smaller than Eve, in order to qualify for the RoboCup competition,
where we entered the 1998 regional contest in Singapore. The original CIIPS
Glory player was equipped with a Color QuickCam camera, which was replaced
in later versions by our own EyeCam design. CIIPS Glory robots have competed
in a number of RoboCup and FIRA World Cup robot soccer events.
We used two servos in addition to the two DC driving motors. These were used
Moving the camera
Kicking the ball
Robot Soccer Team
- The original camera movement was a tilt action, which
allowed us to keep the ball in the (relatively narrow) field of view, when
closing in on the ball. We changed this on later robots in favour of a
panning movement, which allows faster tracking of a moving ball without
having to move the whole robot, or a combination of both.
The goal keepers were a variation of the field player design. Since they
needed to move primarily sideways instead of forward/backward, re mounted
the goal keeper's top plate at a 90 degree angle to the bottom plate and
equipped it with a larger ball kicking plate, as lined out in the RoboCup
Mecanum-wheel design, omni-directional robots, EyeBot controller
- Omni-directional vehicles have a significant advantage
over conventional vehicles with car-like Ackermann steering or the
differential drive using two independent wheel motors as in Eve and many of
our driving robots. Omni-directional driving allows going forward/backward,
but also sideways left/right and turning on the spot. This is especially
helpful when having to maneuver in a tight environment such as a factory
The EyeBot controller drives 4 independent wheels on the omnidirectional
robots Omni-1, Omni-2 and Omni-3. These robots use the "Mecanum" wheel
design with free rollers around the wheel circumference. Each robot can
drive in any direction, i.e. forward/backward, sideways, at an angle, and
turn on the spot. The robots are using the EyeBot controller with an add-on
module with two additional drivers.
- This uses the conventional Mecanum wheel used in Omni-1,
and the suspension system in Omni-2, to create a large scale omni-directional
robot that is used as a wheelchair.
- The conventional Mecanum wheel design with the rollers
held at the sides. This is a disadvantage when driving on non-smooth
surfaces, because the rims will make contact with the surface.
- A new Mecanum wheel design where the rollers are held in
the middle. This gives an advantage when driving on general surfaces,
together with a suspension system that suspends every wheel individually.
Omni-Wheelchair with Driver-Assistance System
Wheelchair with Mecanum wheel omni-directional drive system, sensor-based
driver assistance system for handicapped drivers
- Design Objectives:
- Semi-autonomous omni-directional wheelchair
- Footprint size about 1m x 1m, with payload about 100kg
- Rimmed Mecanum wheel version
- Independent suspension on all four wheels
Tracked driving robots for terrain navigation, Eyebot controller, attitude
and inertial sensors
- The EyeTrack vehicle is a modified model car using
tracks for locomotion. We are using an EyeBot controller for driving the
vehicle and reading its sensor data. Since this robot is to be able to
navigate in terrain, we use a number of orientation sensors to avoid going
up or down too steep inclines. The camera is mounted in an active cardanic
fashion using three servos for three axes.
One of the projects involving EyeTrack is an "intelligent remote control".
The robot drives under remote control and returns images and other sensor
data. However the robot adapts its speed automatically to environment
conditions and refuses to execute any commands that could result in the
robot getting stuck or falling over. This system could be very useful for a
number of rescue or bomb defusing scenarios.
Small size robot soccer team in RoboCup and FIRA WorldCup competitions,
- Our team which has its "home" in the UWA mobile robot
lab was named after the new and successful Perth Glory soccer team. The
heart of the robots are the the EyeBot controllers, developed by a team
around Thomas Braunl. We use a Motorola 68332 32-bit controller, which offer
a variety of digital/analog I/O facilities. We developed our own operating
system RoBIOS for the systems, which allows a great deal of flexibility.
I.e. the same EyeBot system is also used for 6-legged and biped walking
machines, and - as a boxed version - for undergraduate courses in assembly
We incorporated a digital color camera and a graphics display. All image
processing is done on-board. Our robots have local intelligence and are not
simple pawns, directed by a central system with global vision. Although our
approach is clearly disadvanted in respect of winning a RoboCup competition,
we are more interested in research on general purpose intelligent agents, as
opposed to building a system which serves only a certain competition and has
to rely on global sensors.
Each robot has shaft encoders and infrared range sensors in addition to a
digital color camera. We are currently incorporating wireless transmissions
to allow the robots to talk to each other plus a ball kicking device. We are
able to operate even without communication betwen the robots. Each robot
will be told its starting position and will use its shaft encoders to keep
track of its current position. However, communication will allow much more
sophisticated behaviours like passing a ball to another robot.
Biped walking android robots, several generations with different actuator
and sensor equipment
- The mechanics and sensor electronics has been
individually constructed around the EyeBot controller. All robots are about
50cm tall and use different sensors for attitude control and balance.
Andy Droid 2
- The Andy Droid robots have been designed by Joker
Robotics, however, we have modified Andy1's feet with three toes containing
strain gauges as the robot's main orientation sensors. With these, Andy can
always determine the center of pressure in each foot and therefore knows its
"zero moment point" (ZMP), which allows it to counteract any rotational
forces for balancing. Andy2 has an almost identical leg design. Andy2 uses a
new digital servo development instead of Andy1's conventional analog servos.
These digital servos give feedback via a serial interface and can therefore
double as actuators and sensors.
- Johnny and Jack were the first two android robots we
designed using relatively inexpensive servos as actuators and testing a
variety of different sensors. A major problem with these robots are the
total weight of the mechanics, electronics, motors and batteries, and the
rather limited torque being supplied by the motors. Any biped robot that has
difficulties balancing on one leg, will have difficulties to perform a
Another major problem is repeatability of a motion. The robots' metal frame
structure is quite flexible and tends to swing. Also, the inexpensive servos
used have a considerable play and are not capable of an accurate motion
repeat. Furthermore, servo performance significantly decreases with ageing
of the servos.
- With Rock Steady we tried a completely different approach
to our first robots Johnny and Jack (see left). This robot should use a
minimal number of motors: one motor per leg, plus one motor for sideways
balancing of a counterweight. A sophisticated mechanical structure
translates the rotaty motion of each leg motor into an articulated leg
motion. Instead of servos, we used precision DC motors with encoders for
this robot. Inclinometers in two axes are used as orientation sensors for
the robot, each motor has an incremental encoder plus an external zero
position optical switch.
Balancing driving robots as studies for biped sensor equipment, Kalman
- BallyBot is an experimental balancing robot on two
side-by-side wheels, similar to an inverted pendulum. We are using BallyBot
as an experimental platform to gain insight in sensor system and control
systems to be implemented for humanoid robots.
Two of these experimental robots have been built so far. Bally1 (right) is a
simple construction on a single piece of aluminium, holding the controller,
sensor, and two Faulhaber motors with wheels.Bally2 (left) is a more compact
mechanic design with identical sensors, but motors with a higher gear ratio.
The robot is actively balancing and can be driven using IR-Remote input like
a remote controlled car.
Six-legged walking robots with two dof per leg, various sensor equipment
- Walking robots are often slower than driving robots,
but they have the improtant advantage that they navigate over terrain, while
driving robots require a more or less flat surface. The simplest case of a
walking robot uses 6 legs, sinve this allows to implement a gait that always
lifts up and repositions 3 legs, while the other 3 legs remain on the
ground, providing a solid balance. Such a robot does not need to actively
balance, as is required for our balancing and biped walking (android)
robots. For details see the book "Embedded Robotics".
The mechanics and sensor electronics has been individually constructed
around the EyeBot controller.
- 12 servos driven by EyeBot
- 2 infra-red PSD sensors
- mechanics 2nd generation (left) from Lynxmotion
- mechanics 1st generation (right) developed at Univ.
Autonomous submaries / Autonomous underwater vehicles, embedded controller
with inertial sensor system
Underwater Vehicle Projects
- AUV with 4 motors, Eyebot, mini-PC, sensor equipment
- AUV with 2 motors, 1 control surface, Eyebot, sensor
- AUV simulation system