246
Sensemaking in Safety Critical and Complex Situations
This has made it possible to have a more formal definition of constrained autonomy
as provided above.
SUMMARY AND CONCLUSIONS
The concept of constrained autonomy as it has been presented here has been developed over several years and is probably not fully completed yet. However, the concepts and definitions presented in this text are now being used in several development
projects as it is expected that the basic ideas are reasonably stable. There may still be
some changes in the details of the definitions and in how it will be implemented in
actual automation systems. This will be reported on in future publications.
The initial proposal of this text was that a more deterministic automation system
in industrial autonomous systems may increase the operator’s trust in the automation
and by that improve the efficacy of the HAI for periodically unattended autonomous
operations. A central element is to have a verifiable response deadline that can be
matched to the crew’s maximum response time as determined by the organization
of watch-keeping on the ship and in the RCC. The theory is that this is likely to help
in alerting the crew in time to establish a sufficient situational awareness before the
crew is forced to act on situations that the automation system is not able to handle
itself. Thus, the use of constrained autonomy should be a useful way to let the operators make better sense of the interaction between constrained autonomous systems
and the operators. However, the analysis of the human–automation effects is beyond
the scope of this text and will have to be addressed by experts in the human factors field. While the initial proposal seems logical, it will be up to researcher in the
area of human factors to see what actual implications constrained autonomy has on
the operators’ ability to make better sense of the interaction between human and
automation
The main purpose of this text is therefore to describe the technical concept of
constrained autonomy and give it a more formal definition. Some examples of consequences for implementations of automation systems have also been given.
Independent of the h uman-factor angle, it is also believed that the concept can
be used to improve testability and eventually also formal acceptance of autonomous
control systems ( Rødseth 2019). The concept of constrained autonomy may be particularly important for industrial autonomous systems, where the systems are costly,
need to operate in a commercial business model and where the consequences of
system failures may have significant and even catastrophic consequences. Industrial
autonomous systems are also very relevant for the concept of constrained autonomy
as many of them will need an operator in the loop in any case, mainly to oversee the
operation and to safeguard large investments. The examples in this text are from the
maritime domain and the work presented has been focusing on autonomous ships.
However, the concept of constrained autonomy should also be applicable to other
industrial autonomous systems.
The work presented in this text has been partially funded by the Norwegian
Research Council project SAREPTA. It has also received funding from the European
Union’s Horizon 2020 research and innovation programme under grant agreement
No 815012 ( AUTOSHIP).
Sensemaking in Safety Critical and Complex Situations
This has made it possible to have a more formal definition of constrained autonomy
as provided above.
SUMMARY AND CONCLUSIONS
The concept of constrained autonomy as it has been presented here has been developed over several years and is probably not fully completed yet. However, the concepts and definitions presented in this text are now being used in several development
projects as it is expected that the basic ideas are reasonably stable. There may still be
some changes in the details of the definitions and in how it will be implemented in
actual automation systems. This will be reported on in future publications.
The initial proposal of this text was that a more deterministic automation system
in industrial autonomous systems may increase the operator’s trust in the automation
and by that improve the efficacy of the HAI for periodically unattended autonomous
operations. A central element is to have a verifiable response deadline that can be
matched to the crew’s maximum response time as determined by the organization
of watch-keeping on the ship and in the RCC. The theory is that this is likely to help
in alerting the crew in time to establish a sufficient situational awareness before the
crew is forced to act on situations that the automation system is not able to handle
itself. Thus, the use of constrained autonomy should be a useful way to let the operators make better sense of the interaction between constrained autonomous systems
and the operators. However, the analysis of the human–automation effects is beyond
the scope of this text and will have to be addressed by experts in the human factors field. While the initial proposal seems logical, it will be up to researcher in the
area of human factors to see what actual implications constrained autonomy has on
the operators’ ability to make better sense of the interaction between human and
automation
The main purpose of this text is therefore to describe the technical concept of
constrained autonomy and give it a more formal definition. Some examples of consequences for implementations of automation systems have also been given.
Independent of the h uman-factor angle, it is also believed that the concept can
be used to improve testability and eventually also formal acceptance of autonomous
control systems ( Rødseth 2019). The concept of constrained autonomy may be particularly important for industrial autonomous systems, where the systems are costly,
need to operate in a commercial business model and where the consequences of
system failures may have significant and even catastrophic consequences. Industrial
autonomous systems are also very relevant for the concept of constrained autonomy
as many of them will need an operator in the loop in any case, mainly to oversee the
operation and to safeguard large investments. The examples in this text are from the
maritime domain and the work presented has been focusing on autonomous ships.
However, the concept of constrained autonomy should also be applicable to other
industrial autonomous systems.
The work presented in this text has been partially funded by the Norwegian
Research Council project SAREPTA. It has also received funding from the European
Union’s Horizon 2020 research and innovation programme under grant agreement
No 815012 ( AUTOSHIP).