Advanced sensor systems
ET8008
Fall 2009
http://www.hh.se/et8008
Course examiner: Stefan Byttner (Stefan.Byttner@hh.se) Room E505
Lecture schedule:
Thursday 3 sep: Introductory lecture (Stefan Byttner)
Tuesday 8 sep: Camera and vision (Josef Bigun)
Thursday 10 sep: State of the art in physics, nano sensors (Håkan Petterson)
Thursday 17 sep: Virtual sensors for control (Ulf Holmberg)
Tuesday 22 sep: Virtual sensors for monitoring (Antanas Verikas)
http://www.hh.se/schema
press “Schema för flera veckor”. Select “In
english” on the top right. Then you can search
for the course and see the schedule details.
Project work
Deeper study of a particular field of sensors/application. Give some
experience in writing a research report. Possible project examples;
• Review of state of the art of certain sensors
• Experimental evaluation of a sensor setup
• Applications to industrial problems
Course examination:
• Presentation of a related research paper
• Writing of a research report of the project (6 pages or more
depending on format) and oral presentation of the results at a
seminar
Project work (cont.)
Each project will be done by student groups of 2 students
each. Any deviations from this groupsize will be
exceptions and must be approved by examiner. They must
be approved before project work starts.
Seminar with oral presentation of the project work +
presentation of the related paper:
22 Oct 13.15-17 Room D415
Written report sent to course examiner email adress as a
PDF file no later than:
1 Nov 23:59 (Stefan.Byttner@hh.se)
Project work (cont.)
• The report/paper should have a similar type of structure as
the example papers that are handed out from previous year
• It is very important that the report contain some
references/review of earlier (related) work to your chosen
project topic
• In addition to your own work, you need to present a paper at
the seminar. Choose the subject of the paper so that it is
related to your own work (paper choice should be approved
by the examiner once projects have been chosen).
Project work (cont.)
• Fill out the paper which will circulate the class room after the
lecture and decide who will be in each group by yourselves.
• I will email each group with project abstracts, each group then
needs to send me their rank on each project (rank 1 – most
interesting project, rank 2 – second most interesting project,
etc). I will use this for deciding which group gets to do a certain
project. It is possible that there will be more than 1 group
working on the same project.
Grading criterions
• Adopted partly from master thesis grading
• May not need all criterions to be fulfilled to
achieve a certain grade
Grading criterions (cont.)
Information Retrieval
Grade 3
Has compiled a satisfactory list of references.
Knows which references are important and which are less important for the
project.
Can summarize the most important references.
Knows the contents of the remaining references in “abstract” form.
Understands the problem and can formulate subproblems.
Can relate the project to the references.
Grade 4
Can relate the references to each other and combine information from them and
come to conclusions.
Can use the references to sharpen the project plan, i.e. decide to omit some
studies or concentrate on some subproblem.
Grade 5
Is able to criticize and find weak and strong parts in reference articles as well
as the own work.
Grading criterions (cont.)
Results
Grade 3
The result is acceptable, but can point at several things that,
with a reasonable effort, would have improved it.
Can identify and formulate significant strong and weak points in the result.
Grade 4
The result is good (i.e. matches well to the anticipated results in the, possibly
revised, project plan) and can point at only a few things that
could have been done better or that are missing.
Can, with minor supervision, come to some conclusions on how the result
could have been improved.
Can, with minor supervision, formulate some future directions for the project.
Grade 5
The result is excellent and the supervisor can point at only very few minor
improvements which could have been done.
The result is publishable in a scientific journal or at a conference.
Can, without supervision, evaluate the result in relation to other work done
in the field.
Grading criterions (cont.)
Presentation
Grade 3
Presents the problem and proposed solution in a clear way.
Presents a clear analysis of the problem.
Can answer fundamental questions on the subject.
Grade 4
Presents the project in an attractive way with e.g. well chosen
illustrations.
Can discuss different aspects of the problem.
Grade 5
An excellent presentation, which engages the audience and generates
interested questions. All questions are answered in a relevant way.
Grading criterions (cont.)
Report
Grade 3
The report is complete (i.e. with ”background”, ”methodology”,
”results”, ”conclusion”, ”summary” etc.).
All references, figures & tables are referred to in the text.
Grade 4
The report is, with significant supervision, well written. (”Well written” means
that: The English is correct and the text flows smoothly. Figures are relevant
and add value to the text. Similarly with tables. The result and conclusion is
clearly stated.)
Grade 5
The report is, with minor supervision, very well written.
The report is written such that it would be publishable,
provided the result is good enough.
Overview of the sensorbased research at EIS
Research: Physics of Nanoscale Devices
Electrical, optical and magnetic properties of
nanoscale devices
Performed in co-operation with the Nanometer Structure
Consortium at Lund University
1nm
Research: Signal analysis
Speech data
Visual data
Fusion system
• Biometric identification
by combining different
modalities (lip
movements, face,
fingerprints,...)
• Biometric
communication.
Recogniser
Verification results
Reject/Accept
Research: Paper industry
Real time quality measurements (soft sensors)
for paper making
Machine learning, optics, spectroscopy
Shrinking, deinking, …
Research: Print quality
Machine learning and image analysis to measure
color reproduction in printed media. Now for real-time
feedback control
Used by (e.g.) Hallandsposten and Bank of England.
Research: Control theory
Neural network
Ion current
Air/fuel ratio
Pressure
Burn rate
Robust estimation
Robust control
Under the hood of our highway
laboratory, a SAAB 9000 2.3 T
• Real-time on-road system
• DELPHI/GM transfer 99-00
• The best algorithm for this that
DELPHI has tried.
• Product in DELPHI today
Research: Vehicle monitoring
Information communicated
using wireless technology
Vehicle Data
Vehicle Health
Causes for Anomalies :
• Physical Faults
• Operational Abnormalities
• Deterioration
Fleet Health Status
Compare and
Identify
Fleet behavior can provide a
norm in which an “unhealthy”
bus can be detected as
deviating from the norm. The
RDM project investigates how
this norm can be found and
used on a fleet of vehicles to
improve up-time management.
Research: Robotics
Autonomous robots for agriculture (weeding,
sowing, etc.)
Research: Robotics
• Multiple
Autonomous Forklifts for
Loading and
Transportation
Applications
Research: Intelligent home
• An intelligent
environment with
sensors.
• For assisting elderly
people living there.
• Detecting risks of
falling, deviations
from daily pattern of
activities, etc...