09-11-2012, 04:13 PM
DRIVING WITH ADAPTIVE CRUISE CONTROL IN THE REAL WORLD
[attachment=38792]
Abstract
Scientists and researchers still question safety and reliability of various Advanced Driver
Assistance (ADA) systems. One of the central issues remaining is the interaction of ADA
equipped vehicles with other road users. Meanwhile, the first ADA systems have been
introduced into the marked. This paper presents a framework for a study to get towards a
controlled implementation of new or redesigned ADA systems using a systems safety
approach. This approach sets the next step of assessing safety of ADA systems already used
in real world traffic, improving on the currently used limited prospective and retrospective
analyses.
Introduction
Cruise control in vehicles is more and more becoming a standard accessory in modern cars.
To remain distinctive several brands of automotive vehicles are introducing its successor on
the market: Adaptive Cruise Control (ACC). ACC is a system that combines cruise control
with a system for collision avoidance. At first only top models of brands such as Mercedes,
Jaguar, and BMW were equipped with this latest development in automation technology for
automobiles, but recently other brands, such as Nissan, Renault, Fiat, and Volkswagen
introduced ACC in their higher-end models. The system is known under a wide variety of
names, such as: Advance Cruise Control (BMW), Distronic ACC (Mercedes) Intelligent Cruise
Control (Nissan). It is to be expected that middle -class cars will follow in the coming years as
prices of production go down. The same kind of market introduction of systems could also be
witnessed with other devices, such as ABS and in-vehicle navigation systems.
It is normal in the Life Cycle of products that at a certain point during the development of a
new technology there is a moment that market introduction is considered. Even though not
all potential problems with the design have been solved, still there are other criteria that
make market introduction desirable, even unavoidable. This also holds true for ACC. There
are imperfections in the systems but still they have been considered well enough developed
by the car manufacturers to be put on the market.
System safety approach
The examples in the introduction show the diversity of safety aspects that may relate to ACC.
Some of the aspects relate to the proper functioning of the system, while others deal with
the context of the traffic network in which ACC is used by a number of road users. Using a
system safety approach one can distinguish four different system levels at which safety
effects may occur. These levels are: the functional safety level, the driver safety level, the
level of safety of interaction and the traffic safety level. A number of frameworks addressed
the first, second and fourth system level (e.g., Carsten, 1993; Morello, 1995), however a
‘meso’ traffic level at which road users interact with each other in these frameworks is
lacking.
What’s lacking?
Although in a system safety approach problems on four safety levels need to be assessed,
this paper focuses on the issues relating to ‘experiences that differ from pre-introduction
studies’ and ‘unknown and unexpected behaviour of the system and the driver’. These
problems are mainly related to the system levels driver safety and interactions between road
users. In general the problems address attitude and use of ACC at the short and long term.
These four criteria may be further specified for a particular evaluation. For our discussion we
are interested to what extend these criteria have been addressed and what kinds of methods
were used to do so.
In a limited literature review a total of 20 evaluation studies going into aspects of attitude or
use of Adaptive Cruise Control (ACC) or Intelligent Speed Adaptation (ISA) were selected
(Bianchi, 2003). Different from ACC, ISA has not yet been introduced to the market,
although a number of manufactures provide an option to set one maximum speed limit into
the board computer of their vehicles. ISA has been added to the literature review as in
recent years a number of experiments with ISA in real traffic conditions were performed.
Table 1 shows the methods that were used to address attitude towards ACC or ISA or use of
ACC or ISA. As most of the studies describe more than one of the criteria, the total number
of entries in the table is lager than the number of selected studies. The table shows that the
evaluation studies mainly address criteria on the short term. Most of the studies found in
literature relate to testing of the intended use and status of the systems. However, a lot of
safety problems relate to deviations of such intended use and status. The problem we
experienced during our test drive with a third vehicle changing lanes to exit the motorway is
an example of such a deviation. If we fill in Table 1 only for problems relating to unknown
and unexpected behaviour of the system and the driver most cells will be empty. This is
mainly caused by the lack of an approach to identify such possible deviations before to set
up the scenarios for a safety assessment. The use of hazard and operability studies (HAZOP)
could be used to for defining these scenarios (see, Jagtman, 2002)
Discussion
Ideally, one would perform a number of experiments to study the attitude and use of drivers
on the short term at early stages of development of ADAS followed by an extensive field test
period long enough to get knowledge on possible changes in attitude and use after the
learning phase has passed. However, it is questionable if the expenses for such long term
field trials can brought in with the increasing number of such systems that are expected to
come to the market. Secondly, it is impossible to foresee all possible consequences of the
use of ACC in the real world in advance, either positive or negative. At the introduction of a
system there will always be remaining uncertainties. At some point however these
uncertainties may be qualified as limited to an acceptable level for market introduction.
[attachment=38792]
Abstract
Scientists and researchers still question safety and reliability of various Advanced Driver
Assistance (ADA) systems. One of the central issues remaining is the interaction of ADA
equipped vehicles with other road users. Meanwhile, the first ADA systems have been
introduced into the marked. This paper presents a framework for a study to get towards a
controlled implementation of new or redesigned ADA systems using a systems safety
approach. This approach sets the next step of assessing safety of ADA systems already used
in real world traffic, improving on the currently used limited prospective and retrospective
analyses.
Introduction
Cruise control in vehicles is more and more becoming a standard accessory in modern cars.
To remain distinctive several brands of automotive vehicles are introducing its successor on
the market: Adaptive Cruise Control (ACC). ACC is a system that combines cruise control
with a system for collision avoidance. At first only top models of brands such as Mercedes,
Jaguar, and BMW were equipped with this latest development in automation technology for
automobiles, but recently other brands, such as Nissan, Renault, Fiat, and Volkswagen
introduced ACC in their higher-end models. The system is known under a wide variety of
names, such as: Advance Cruise Control (BMW), Distronic ACC (Mercedes) Intelligent Cruise
Control (Nissan). It is to be expected that middle -class cars will follow in the coming years as
prices of production go down. The same kind of market introduction of systems could also be
witnessed with other devices, such as ABS and in-vehicle navigation systems.
It is normal in the Life Cycle of products that at a certain point during the development of a
new technology there is a moment that market introduction is considered. Even though not
all potential problems with the design have been solved, still there are other criteria that
make market introduction desirable, even unavoidable. This also holds true for ACC. There
are imperfections in the systems but still they have been considered well enough developed
by the car manufacturers to be put on the market.
System safety approach
The examples in the introduction show the diversity of safety aspects that may relate to ACC.
Some of the aspects relate to the proper functioning of the system, while others deal with
the context of the traffic network in which ACC is used by a number of road users. Using a
system safety approach one can distinguish four different system levels at which safety
effects may occur. These levels are: the functional safety level, the driver safety level, the
level of safety of interaction and the traffic safety level. A number of frameworks addressed
the first, second and fourth system level (e.g., Carsten, 1993; Morello, 1995), however a
‘meso’ traffic level at which road users interact with each other in these frameworks is
lacking.
What’s lacking?
Although in a system safety approach problems on four safety levels need to be assessed,
this paper focuses on the issues relating to ‘experiences that differ from pre-introduction
studies’ and ‘unknown and unexpected behaviour of the system and the driver’. These
problems are mainly related to the system levels driver safety and interactions between road
users. In general the problems address attitude and use of ACC at the short and long term.
These four criteria may be further specified for a particular evaluation. For our discussion we
are interested to what extend these criteria have been addressed and what kinds of methods
were used to do so.
In a limited literature review a total of 20 evaluation studies going into aspects of attitude or
use of Adaptive Cruise Control (ACC) or Intelligent Speed Adaptation (ISA) were selected
(Bianchi, 2003). Different from ACC, ISA has not yet been introduced to the market,
although a number of manufactures provide an option to set one maximum speed limit into
the board computer of their vehicles. ISA has been added to the literature review as in
recent years a number of experiments with ISA in real traffic conditions were performed.
Table 1 shows the methods that were used to address attitude towards ACC or ISA or use of
ACC or ISA. As most of the studies describe more than one of the criteria, the total number
of entries in the table is lager than the number of selected studies. The table shows that the
evaluation studies mainly address criteria on the short term. Most of the studies found in
literature relate to testing of the intended use and status of the systems. However, a lot of
safety problems relate to deviations of such intended use and status. The problem we
experienced during our test drive with a third vehicle changing lanes to exit the motorway is
an example of such a deviation. If we fill in Table 1 only for problems relating to unknown
and unexpected behaviour of the system and the driver most cells will be empty. This is
mainly caused by the lack of an approach to identify such possible deviations before to set
up the scenarios for a safety assessment. The use of hazard and operability studies (HAZOP)
could be used to for defining these scenarios (see, Jagtman, 2002)
Discussion
Ideally, one would perform a number of experiments to study the attitude and use of drivers
on the short term at early stages of development of ADAS followed by an extensive field test
period long enough to get knowledge on possible changes in attitude and use after the
learning phase has passed. However, it is questionable if the expenses for such long term
field trials can brought in with the increasing number of such systems that are expected to
come to the market. Secondly, it is impossible to foresee all possible consequences of the
use of ACC in the real world in advance, either positive or negative. At the introduction of a
system there will always be remaining uncertainties. At some point however these
uncertainties may be qualified as limited to an acceptable level for market introduction.