The Intelligent Servosystems Laboratory
P. S. KrishnaprasadDepartment of Electrical and Computer Engineering
Institute for Systems Research (ISR)
University of Maryland, College Park
July 25, 2003
INSTITUTE FOR SYSTEMS RESEARCH
BRIEF HISTORY
• 1986: Began as a small lab devoted to the study of– Control of flexible-link robotic arms
(Westinghouse)– Tactile sensory processing for Si membrane
sensor (NRL)– 3-D graphics and related algorithms (NASA,
AFOSR)
BRIEF HISTORY (cont’d)
• Evolution in A.V. Williams Bldg. as a lab devoted to the study of – actuators– manipulators– complex multi-body systems
via physical experiments and simulations.
BRIEF HISTORY (cont’d)
• Highlights include experiments in– friction modeling and adaptive control
(Westinghouse)– impact control (ARO)– design, fabrication and grasp analysis (NSF)– testbed for space robotics (NASA)– smart motor network (NSF)– walking robots (NASA, NSF)
Modular dextrous hand, (NSF, AFOSR)Loncaric, de-Comarmond, ...
Hybrid motor (NSF, ARO), Venkataraman,Loncaric, Dayawansa,Krishnaprasad
Parallel manipulator (NASA, NSF, DOE)Tsai, Tahmasebi, Stamper,...
Snakeboard
Video here
Roller Racer MoviesSean -straight line motion
Sameer - circular arcsand figure eight
Paramecium
Joseph-Louis Lagrange Jean Le Rond d’Alembert Emmy Amalie Noether (1736-1813) (1717-1783) (1882-1935)
GOALS OF LAB
• To advance the state-of-the-art in design and real-time control of smart systems, drawing on advances in – Novel sensing and actuation materials and
mechanism designs– New principles for actuation, propulsion,
detection, reduction, learning and adaptation– Conceptualizing and prototyping across scales, to
sense, actuate, communicate and control
RESEARCH & EDUCATION
The lab as a facility for education and training – Education and training of some 30 M.S.
students and 34 Ph.D. students since 1986, all of whom had participated in some significant way in the lab through experimental and computational investigations in addition to engaging in theoretical investigations.
RESEARCH & EDUCATION
The lab as a facility for education and training – Involvement of undergraduates and high school
students in the lab through REU programs and Young Scholar programs, continually, over the past 14 years.
ROLE IN ISRFocus on
• research in problems of interaction of physical systems with software systems– mobile robots with on-the-fly motion planning
algorithms
• integrated approach to the design and control of smart systems– biologically inspired approaches to sensory
processing and motion control
ROLE IN ISRFocus on• research in the science of controllable
materials– hysteresis from first principles– nonlinearity in adaptive optics
• A facility for research in real time control prototyping– dSPACE
Smart System
Neuroscience
Materials
Intelligent Control
Modeling & Optimization
Wireless
Noise &Sensors
Smart PowerMEMS
Signal Processing
Robotics
LINKAGES
CURRENT PROJECTS
• Robotics (serial, parallel, mobile, small, …)– GPS-enhanced robotics
• Smart materials, devices and systems (CDCSS)– Hysteresis– Actuator arrays for adaptive optics (CDCSS
and ARL)
CURRENT PROJECTS
• Smart manufacturing (NG, NSF)– Understand Si epitaxy, Si-Ge epitaxy via CFD– Process Control
• Links with biology (LIS, CAAR, NACS)– Learning and intelligence– Robotic barn-owl
ROBOTIC BARN OWL
GPS CAR
GPS CAR
GPS CAR AND ANTENNA
MENTORING
GPS CONFIGURATION
GPS is a satellite navigation system using NAVSTAR satellitesTwenty-four operational satellites provide GPS receivers with satellite coverage at all times
GPS Orbit Configuration
GPS-aided Location Determination and Navigation
Receiver requires minimum of four satellites
GPS in Differential mode
EquationsDistance from receiver to satellite given by:
Pik = i
k + c [dti - dtk] + Tik + Ii
k + dik + ei
k
Pik = pseudorange
ik = ||ri – rk|| = distance between satellite and
receiver
= {(Xi – Xk)2 + (Yi – Yk)2 + (Zi – Zk)2}1/2
ri = position of receiver
rk = position of satellite
c = speed of light
dti = clock bias in receiver
dtk = clock bias in satellite
Tik = Tropospheric correction
Iik = Ionospheric correction
dik = multipath correction
eik = noise
Linearized form of Equation (Using Taylor series)
Pi = -(Uik) ri + Cdti + i
k
Pi = observed position – calculated position
(Uik) = unit vector from receiver to satellite
ri = actual position – initial estimate
dti = actual receiver clock bias – initial estimate
ik = error terms
Global to Local coordinates transformation
-Sini*Cosi Sini -Cosi*Cosi
ni = -Sini*Sini ei = Cosi uI = -Cosi*Sini
-Cosi 0 -Sini