This document discusses the ironies of automation and opportunities for human-machine interactions. It notes that while automation has benefits, it can also dull human senses, making operators less able to notice problems. The key is designing systems to accommodate and support human factors rather than substitute for them. It proposes using adaptive automation and data models to better understand humans and build close feedback loops between user actions and system effects. The goal is to leverage the best of human intuition and machine data processing to improve overall performance.
3. Theironiesofautomation01
“Ironically, reliable systems tend to dull the operator’s senses, making it much harder
for them to notice when things go wrong”
“the more advanced a control system is, the more crucial may be the
contribution of the human operator”
6. Why Human Factors are so critical
in system design?
04
Humans are what keep incomplete system working.
The problem is that man often appears to be the weak link of the chain and
the more unpredictable:
Therefore we try to design to substitute the HF rather than accommodate
and support it
7. Brief overview of the state of the art05
Quantitative Qualitative
General
Methods
Very
specific
methods
Static Dynamic
Quantitative
man-system
interaction
Generalizable
FIRST GENERATION
METHODS:
COGNTIVE SIMULATION
(SECOND GENERTION)
Common problem for both first and second
generation HRA approaches:
data availability
8. Nature of the Problem06
• Dangerous activities are evaluated in terms of probabilities of accident
occurrences
• Being human failure among the main accident causes it is often
required to be part of the risk assessment framework in which plays an
important role
First generation Rough Engineer approach:
human reliability = hardware reliability
Second generation Human Scientist approach:
human reliability should be treated as a concept not as a number
(Hollnagel 1991)
9. An Updatable data model BBN07
Legend:
Human
Performance
Task
Complexity
Human
Capability
Physical
Workload
Physical
ability
Mental ability
Mental
Workload
O
B
S
E
R
V
A
B
L
E
V
A
R
I
A
B
L
E
S
Best matching
HMI
support
Variables to be
measured
Derived
Variables
Individual
Motivation
Job characteristic
motivational score
10. Critical Elements to consider08
• Performance shaping factors: the role of time, task complexity environmental
conditions.
• The connection between the human actions and the plant/work environment
conditions
• Mathematical model of the interaction.. Updatable data model
• Verification of the robustness of the main hypothesis
11. The benefits of adaptive automation:
new possibilities
09
“Humans are doing a pretty good job, but they do it even
better with the assistance of algorithms… when algorithms
work with humans, the whole system performs better.”
Mary L. Cummings, Director of the Humans and
Automation Laboratory, at (MIT) and a former
Navy F-18 pilot
12. A different type of Turing test?10
Can we use AI to better support the human?
13. NEW HORIZONS: KEY INGREDIENTS11
1. Understand better the human., data models more tuned in into
realistic representations
2. Use advances in Neuroergonomics for real time detection of
changes in our conditions
3. Support situational awareness …build close feedback loop
between actions and effects,
4. Use the best of both worlds (fast data processing, for MACHINE
, power of intuition for THE HUMAN
14. Conclusion12
“They constantly try to escape
From the darkness outside and within
By dreaming of systems so perfect that no one will
need to be good.
But the man that is shall shadow The man that
pretends to be.“
T.S. Eliott
Performance is the ultimate product of the
balance between task complexity and capability,
we want to use the system to improve our
strenghts and mitigate our wekannesses