Transcript Slide 1

UNIVERSITY OF AVEIRO
Centre for Mechanical Technology and Automation
Institute of Electronic Engineering and Telematics
Engineering Solutions to Build an Inexpensive Humanoid
Robot Based on a Distributed Control Architecture
1
Vítor M.F. Santos and Filipe M.T. Silva
2
1
2
Department of Mechanical Engineering, University of Aveiro, PORTUGAL  [email protected]
Department of Electronics and Telecommunications, University of Aveiro, PORTUGAL  [email protected]
Introduction
Mechanical Design
Head
base
6 DOFs per leg
 Objectives/Motivation
 Develop a humanoid platform for research on control,
navigation and perception
 Offer opportunities for under & pos-graduate students to
apply engineering methods and techniques
 Build a low-cost humanoid robot using off-the-shelf
technologies, but still aiming at a fully autonomous platform
 Spherical joint on the hip
 Simple joint on the knee
 Universal joint at the foot
Neck & head
accounts for 2
DOFs
Upper hip
3 DOFs
per arm
Lower hip
Neck
Ankle
Foot
 Why not a commercial platform?
 Versatile platforms imply prohibitive costs!
 Reduces involvement at lowest levels of machine design
 Design Concerns
 Distributed control architecture due to the complexity
 Modularity to ease development & scalability
 Rich sensorial capabilities
Upper leg
Shoulder
Trunk
FINAL PLATFORM
3D model with
600+ components
and 22 DOFs
Lower leg
Pelvis
Hip
Trunk with
2 DOFs
Actuators
Forearm
Power Requirements
 Static-dynamic simulations were carried
out to estimate motor torques
 Motors
 Max current: 1.2–1.5 A per motor
(big size model)
 Best low cost actuators in the market
are Futaba RF servos or similar (HITEC,…)
 Electronics and control
 Available models best suited for our
application are:
Application
Model
Arms & small torque joints
HS85BB
~20
0.35
HS805BB
119
2.26
Legs & high torque joints
Arm
 Estimated to less than 200 mA
per board  a total of ca. 1.5 A
 Voltage Levels
Mass (g) Torque (Nm)
 5 V for logic; 6.5 V for motors
 Two ion-lithium batteries
installed  from Maxx Prod
 7.2 V/9600 mAh per pack
 Max sustained current of 19A
 Each weights circa 176g
 Confined to 373765 (mm3)
 Additional mechanical issues for motors
 Use gear ratios up to 1:2.5 to rise
torques
 Use tooth belt systems for easier tuning
 Use ball bearings and copper sleeves
Sensorial Capabilities
Vision unit
(on the head)
Sensitive Force
Gyroscopes for
angular velocity
GYROSTAR
ENJ03JA from
MURATA
Potentiometer for
position feedback
(HITEC Motor)
 A device was custom-made using strain gauges
properly calibrated and electrically conditioned
 Four strain gauges arranged near the corners of the foot
Accelerometers for
accelerations/inclinations
ADXL202E from
ANALOG DEVICE
Motor electric current
Serial power
resistor
Sensitive feet
Strain gauges on a slightly
compliant material
Adjustable screw
Strain Gauge
Flexible beam
Foot base
UNIVERSITY OF AVEIRO
Centre for Mechanical Technology and Automation
Institute of Electronic Engineering and Telematics
Engineering Solutions to Build an Inexpensive Humanoid
Robot Based on a Distributed Control Architecture
1
Vítor M.F. Santos and Filipe M.T. Silva
2
1
2
Department of Mechanical Engineering, University of Aveiro, PORTUGAL
Department of Electronics and Telecommunications, University of Aveiro, PORTUGAL
Control Architecture
Control Level Units
Main Control
 Distributed control system
RS232
 A network of controllers connected
by a CAN bus
 A master/multi-slave arrangement
 Each slave controller is made of a PIC
device with I/O interfacing
Master
CAN
BUS
Slaves
3
1
3
2
2
2
2
1
1
1
 Asynchronous communications
 Between master and slaves: CAN bus
at 1 Mbit/s
 Between master and high level
controller (currently serial RS232 at
3
3
2
2
1
1
3
3
2
2
1
1
 Main control unit
 Global motion directives & high level planning
 Vision processing
 Interface with possible remote hosts
 1 Master CAN controller
 Receives orders to dispatch to the slaves
 Queries continuously the slaves and keeps the sensorial
status of the robot  currently does it at ca. 10 kHz
 7 Slave CAN controllers
 Generate PWM for up to 3 motors
 Interface local sensors
 Can have local control algorithms
38400 baud)
Local Control Boards
Low Cost… How Low?
 Servomotors
 Master and slave boards have a common base
upon which a piggy-back unit can add I/O 
 Big size: ~50 € x 14 -> 700 €
 Smaller size: ~30 € x 8 -> 240 €
sensors, communications …
Power regulator
PIC
 Miscellaneous electronic components
Reset button
 Total -> ~300 €
Power plug
PIC Cristal
oscillator
Power resistor
(0.47)
CAN bus
Piggy-back
board 1
 Aluminium gears and belts
PWM plugs
CAN
connector
Piggy-back
socket
 Total -> ~300 €
RS232
plug
Servo fuse
Fuse
status LED
16:1
multiplexer
CAN driver
 Batteries
Unit CAN
Address
Piggy-back
board 2
Connector
to sensor
 ~80 € x 4 -> ~320€
Connector
to sensor
 Sensors (except camera)
 Negligible (<100€)
 PIGGY-BACK BOARDS
 Accelerometers
 Serial COM
for master
 Raw materials (steel, aluminium)
 Strain gauges
conditioning
 Negligible (<100€)
 Total ~ €2000
 Excluding manufacturing and
development costs (software, etc.)
Dual amplifier
 Still missing:
Dual
accelerometer
 Vision unit, central control unit (PC104+),
lots of software
25 mm
Ongoing/Open Issues
Conclusions
 First humanoid motions
 The robot is able to stand, lean on sides/forward/backward
 Primitives locomotion motions have been achieved
 The distributed architecture shows several benefits:
 Next concerns for the platform
 Joint position feedback from dedicated sensor
(not servo’s own!)
 Safety issues to automatic cut of power on controller failure
 Better adjustable tensors for belts
 Selection and installation of central control unit
(Embedded
Linux)
 Selection and installation of the vision unit
 Research concerns
 A highly versatile platform is possible to be built with
constrained costs and off-the-shelf components
(FireWire..?)
 Easier development
 Easier debugging
 Modular approaches
 Localized control
 The
selected
 local controllers using piggy-back modules
 based on local perception and global directives
technological
platform for research, mainly on:
 Control algorithms
 Localized/distributed control algorithms
 Perception
 Elementary gait definition
 Autonomous navigation
…
solutions
ensure
a