city college of new york 1 john (jizhong) xiao robotics and intelligent systems lab department of...
TRANSCRIPT
1City College of New York
John (Jizhong) XiaoRobotics and Intelligent Systems LabDepartment of Electrical Engineering
City College of New York
Tel: 212-650-7268Email: [email protected]: http://134.74.16.73
Robotics Research at CCNY
2City College of New York
Current Projects• NSF MII Planning Project:
– Center of Perceptual Robotics and Intelligent Systems (PRISM Center at CCNY)
– Funding Agency: NSF Minority Institutional Infrastructure Program
• Wall-climbing Robot project – Funding Agency: Army Research Office
• Smart Brain project – Funding Agency: NSF Major Research Instrument
Program
3City College of New York
Wall-climbing Robot Project• Project Title:
– Cooperative Wall-climbing Robots in 3D Environments for Surveillance and Target Tracking
• The objective: – develop a modular, re-configurable, wall-climbing
robotic system and to investigate intelligent control methods and vision algorithms to control and coordinate a team of such robots to perform various defense, security, and inspection missions.
• Principle Investigators:– John Xiao (EE), Zhigang Zhu (CS)– Ali Sadegh (ME)
4City College of New York
Wall-climbing Robot Project• Dream:
– transform the present 2-D world of mobile rovers into a new 3-D universe.
– move on ground, climb walls, walk on ceilings, transit between surfaces.
• Applications:– Urban warfare applications: surveillance and reconnaissance,
weapon delivery, guiding perimeter around a building, etc– Security and counter-terrorist applications: intelligence
gathering about a hostile situation within a building, etc.– Inspection and maintenance applications: routine inspection
of buildings, nuclear containment domes, and other hard-to-reach places, inspection of aircraft, sand blasting of ship hulls, etc.
– Other Civilian applications: assistance in firefighting, search and rescue operations, etc.
5City College of New York
Wall-climbing Robot Project• Challenges:
– Adhesive mechanism • strong attraction force on various wall surfaces
(brick, wood, glass, stucco, plaster, and metal) • without sacrificing mobility
– Transition Mechanism• wheeled robot to achieve quick motion• articulated structure for smooth transition• modular design combine two
– Control/coordination of multiple robot modules– Vision research for surveillance applications
6City College of New York
Existing Technologies and Robots
• Existing Technologies– magnetic attraction devices– vacuum suction techniques– biologically inspired
• gecko foot • limbed devices
– aerodynamics attraction• vortex attraction technique• attraction generated by propeller
7City College of New York
Existing Technologies and Robots• Existing wall-climbers:
MSU “Flipper” & “Crawler”JPL-Stanford rock climber
Avionic Instruments Inc. Vortex attraction technique iRobot’s Mecho-Gecko
CM
U gecko inspired
climber
8City College of New York
Wall Climber: Adhesive Mechanism
• Design alternatives:– vacuum pumps (MSU climber)– vortex attraction device– vacuum rotor package
Vacuummotor
air out
air out
air in
Vacuumimpeller
Exhaust(outer)
Exhaust(inner)
9City College of New York
Wall Climber: Vacuum Chamber Seal
• Inflated Tube Skirt Seal
• Flexible Bristle Skirt Seal
attraction force is so strong that it anchored the device to wall surfaces
trade-off between sealing and mobility
10City College of New York
Wall Climber: Selected Design
• Selected Design– vacuum rotor package– flexible bristle skirt seal– differential drive – pressure force isolation rim (re-foam)
• improves mobility, & enhances sealing by reducing the deformation of the skirt
R e-fo am
S k irt
V acu u m O ff
R e-fo am
S k irt
V acu u m O n
R eac tio n fo rce s fro m w e ig h tan d p re ssu re fo rce o n
o u te r r im a rea o n ly
R eac tio n fo rce s fro m w e ig h tan d p re ssu re fo rce o n
o u te r r im a rea o n ly
Plate
R e-foam
P ressu reF o rce
D rive wheel
R eac tio n fo rce s fro m w e ig h t
11City College of New York
Wall Climber: Transition Mechanism
• Modular Design
Four wall-climber modules are configured to form a larger wall-climbing robot which can carry heavy payload
• Transition Mechanism
12City College of New York
DSP-based Control System
33887M otoro la
F2812 D SP
33887M otoro la
M 2
M 3
M 1
O UT1
O UT2
O UT2
IN1
IN2
IN1
IN2
PW M1
PW M2
PW M3
PW M4
PW M5
PW M6
PW M7
PW M8
M 1Encoder
M 2Encoder
M 3Encoder
C hA
C hA
C hA
C hB
C hB
C hB
Q EP1
Q EP2
Q EP3
Q EP4
CAP3
CAP6
ADCINB4
ADCINB3P-Sensor2
P-Sensor1
PresssureSensor
G PIO B6Valve1
SC I-B
G PIO F, 8,9,10,11,12,13
6 D igita l I/O Sensors
G PIO B2 EN
G PIO B3 EN
G PIO B4
G PIO B5
R S232 R eceiver
IRSensor
(SH AR P)
ADCINA7ADCINB5ADCINB6ADCINB7
O UT1
33887M otoro laO UT2
IN1
IN2
ENO UT1
33887M otoro laO UT2
IN1
IN2
EN
O UT1XINT1
G PIO B7
FB
FB
FB
M 3
D rive M otor
D ecoder
U ltrason icS ensor
eco
Trig
M A R G
M agnetic
Accelerom eter
G YR O
ADCINB0ADCINB1ADCINB2
ADCINA3
ADCINA0ADCINA1ADCINA2
ADCINA4ADCINA5ADCINA6
R S232H ost C om puter
SC I-A
D rive M otor
L ift M otor
Vacuum M otor
• Actuator and sensor suite
•TMS320F2812 DSP from Texas Instruments Inc.
• 32-bit Processor
• Target for control applications
13City College of New York
Wall Climber: Software Structure
M otorController
Task LevelScheduler
TrajectoryP lanner
M otor1
Encoder
M otor 2
Encoder
Encoder
Valve
C lim bing R obot
P ressure S ensors
M otion S tatus
Desired angles
Task leve lcom m ands
M otion steps
InfraredSensor
SonarSensor
M ARGsensor
M icrocam era
M otor 3
Motor
Com m andInterpreter
M otionP lanner
Rem oteContro ller
O peratorCom m ands
Feasib le m otionsequence
In itia l & goa lconfigura tions
E nvironm ent
W all Surface
14City College of New York
CCNY Wall Climber Prototypes
Prototype III, vacuum rotor package
Prototype II, inflated tube seal
Prototype II, flexible bristle skirt seal
Prototype I, vortex attraction
15City College of New York
CCNY Wall Climber Prototypes
• Video
16City College of New York
Smart Brain Project• Project Title:
– Smart Re-configureable Miniature Robot Systems Based on System on Programmable Chip Technology
– NSF MRI Instrument Development
• The objective:– to develop highly-adaptive computation module based
on SoPC technology (FPGA) for ultra-small robots – to realize onboard sensor processing, advanced motion
control, and reliable wireless communication
• Principle Investigators:– Umit Uyar (EE), John Xiao (EE)
17City College of New York
Project Overview• FPGA technology
– programmable logic programmable systems– integrate FPGA logic, embedded high-performance
processors, digital signal processor (DSP) blocks, and multi-gigabit transceivers, making FPGA a versatile technology for high-end research and commercial products.
• FPGA Device Features– Xilinx Virtex-II Pro family FPGA device – two 32-bit IBM PowerPC 405 cores– FPGA logic, DSP blocks– 10M of block RAM, off-chip memory as a gap-stop
measure – Xilinx Intellectual Property (IP) Core library
18City College of New York
Project Overview
• Benefits– flexibility, reconfigureability
• hardware reconfigureable, software reprogrammable
– hardware/software partitioning • high-speed logic implementation in FPGA fabric &
high-flexibility software code in Power PC
– IP core library to achieve basic robotic functions• pre-verified, reusable• satisfy the requirements for control, communication,
and onboard vision processing capability of miniature robots
19City College of New York
FPGA-based Multiprocessor
O n-C hip Periphera l BU S (O PB)
PLBArbiter
Pro
cess
or
Lo
cal B
US
(P
LB
)
O PBArbiter
PLB to O PBBridge
PLB B lockR AM
InterfaceC ontro ler
PLB B lockR AM
InterfaceC ontro ler
IIC BusInterface
D iscrete C osineTransform
(2D D C T V2.0)
Fast FourierTransform(FFT V2.0)
PLBArbiter
PLB to O PBBridge
O n-C hip Periphera l BU S (O PB)O PB
Arbiter
IIC BusInterface
G PIO
SD R AMC ontro llerO PB
Tim er/C ounter
PLB B lockR AM
InterfaceC ontro ler
Pro
cess
or
Lo
cal B
US
(P
LB
)
D M AC ontro ller
D ual PortBR AM(V5.0)1A0
1A16
2A0
2A16
2D0
2D15
1D0
1D15
2O E
1W E 2W E
1O E
BinaryC ounter(V6.0)
CLK
ACLR
Q 0
Q 16
PowerPC405 Core
PowerPC405 Core
PW M
Q uadratureLogic
M otor
PowerAm plifier
Phase A
Phase B
PW MLogic
Color Cam eraM odule
SD A
SC L
Y0-Y7U V0-U V7
H R EFPC LK
VSYN C
16
17
8
8
Virtex-II ProFPGA
U AR TRF
M odule
Contro ller
Enc
oder
A/D
Sensors
Auto-ScanC ircuitry
C2I
O PBTim er/
C ounter
CAPT0 CAPT1
INT
D ual PortBR AM(V5.0)1A0
1A16
2A0
2A16
2D0
2D15
1D0
1D15
2O E
1W E 2W E
1O E
EncoderCounter
• A processor-centric architecture
• FPGA fabric is used for custom logic and interfaces.
• Single board FPGA-based multiprocessor for robotics applications
20City College of New York
Recent Progress
• Logic design for motor control (PWM, encoder reading)
• Demonstration of virtual backbone concept for reliable server pooling
Virtex-II Pro ML300 Evaluation board
21City College of New York
Thank you!