Distributed Adaptive Control: An Ideal Cognitive Architecture Candidate
155
modelling the human mind. Thus, systems driven by a cognitive architecture could reason about problems across different domains, develop insights, adapt to new situations
and reflect on themselves [9].
Taking advantage from such cognitive architecture working in a recursive fashion,
we propose the creation of a factory operating as a synthetic agent. The term recursive
refers to the double-scale functionality of the system: various individual agents that
are specialised in different tasks and that are controlled by a higher-level entity, the
factory itself. Envisioning a recycling plant with a recursive architecture resonates with
the metaphor “Der Mensch als Industriepalast” (Man as an industrial palace) that was
proposed by Fritz Kahn. Analogous to a recycling plant, the ingestion of food in Fritz’s
illustration implies a procedure of treatment, disassembling and classification of the
material into different nutrients. For a seamless usage of this disassembled material,
the control centre (in the representation of the brain) works on top of the other related
systems such as metabolism, blood circulation or respiration.
However, Fritz’s illustration depicts a linear process where individual agents are not
required to perform dynamically and are hence not involved in the implementation of a
cognitive architecture. In contrast, the recycling plant presented here requires a cognitive
architecture at both single-agent and large-scale levels, since robots performing tasks of
navigation, disassembling, classification, etc. need to work autonomously in a parallel
and context-adaptative fashion. The synergic operation of the whole plant depends on the
agents’ behaviours, which are monitored, controlled and influenced by the large-scale
level.
In the following sections, we propose the Distributed Adaptive Control theory of
mind and brain (DAC) as a candidate to control a hybrid human-robot recycling plant
of Waste Electrical and Electronic Equipment (WEEE) management. Previous work
on DAC will elucidate how this architecture supports essential robots’ abilities at both
single-agent and large-scale levels. Finally, we present a micro-recycling plant as a
functional prototype and a benchmark for the implementation of DAC.
2 WEEE Recycling Plant
Recycling awareness is gaining importance, especially since recycling plants have to
deal with a significant amount of waste per year. The European Union by 2017 recycled
and composted 94 Mt (35.2%) of its municipal solid waste [10]. For this reason, and
as in many other industries, recycling plants have incorporated a variety of machinery
which processes paper, glass, plastics and other materials on a large scale.
However, society is consuming a growing number of electrical and electronic devices
that, after a few years, become into e-waste. This e-waste, namely Waste Electrical and
Electronic Equipment (WEEE), cannot be processed by the machinery designed to handle the raw materials mentioned previously. The challenge in WEEE management does
not only lie in the correct classification of a device but also its disassembly. Each device
(such as a TV screen) has a variety of models, and not all models include the same disassembly procedure, or each procedure may be executed in a different order. Additionally,
the handling of sensitive or hazardous material (like mercury lamps) adds an extra level
of difficulty in the automation of the processing of e-waste. So far, humans seem to be
155
modelling the human mind. Thus, systems driven by a cognitive architecture could reason about problems across different domains, develop insights, adapt to new situations
and reflect on themselves [9].
Taking advantage from such cognitive architecture working in a recursive fashion,
we propose the creation of a factory operating as a synthetic agent. The term recursive
refers to the double-scale functionality of the system: various individual agents that
are specialised in different tasks and that are controlled by a higher-level entity, the
factory itself. Envisioning a recycling plant with a recursive architecture resonates with
the metaphor “Der Mensch als Industriepalast” (Man as an industrial palace) that was
proposed by Fritz Kahn. Analogous to a recycling plant, the ingestion of food in Fritz’s
illustration implies a procedure of treatment, disassembling and classification of the
material into different nutrients. For a seamless usage of this disassembled material,
the control centre (in the representation of the brain) works on top of the other related
systems such as metabolism, blood circulation or respiration.
However, Fritz’s illustration depicts a linear process where individual agents are not
required to perform dynamically and are hence not involved in the implementation of a
cognitive architecture. In contrast, the recycling plant presented here requires a cognitive
architecture at both single-agent and large-scale levels, since robots performing tasks of
navigation, disassembling, classification, etc. need to work autonomously in a parallel
and context-adaptative fashion. The synergic operation of the whole plant depends on the
agents’ behaviours, which are monitored, controlled and influenced by the large-scale
level.
In the following sections, we propose the Distributed Adaptive Control theory of
mind and brain (DAC) as a candidate to control a hybrid human-robot recycling plant
of Waste Electrical and Electronic Equipment (WEEE) management. Previous work
on DAC will elucidate how this architecture supports essential robots’ abilities at both
single-agent and large-scale levels. Finally, we present a micro-recycling plant as a
functional prototype and a benchmark for the implementation of DAC.
2 WEEE Recycling Plant
Recycling awareness is gaining importance, especially since recycling plants have to
deal with a significant amount of waste per year. The European Union by 2017 recycled
and composted 94 Mt (35.2%) of its municipal solid waste [10]. For this reason, and
as in many other industries, recycling plants have incorporated a variety of machinery
which processes paper, glass, plastics and other materials on a large scale.
However, society is consuming a growing number of electrical and electronic devices
that, after a few years, become into e-waste. This e-waste, namely Waste Electrical and
Electronic Equipment (WEEE), cannot be processed by the machinery designed to handle the raw materials mentioned previously. The challenge in WEEE management does
not only lie in the correct classification of a device but also its disassembly. Each device
(such as a TV screen) has a variety of models, and not all models include the same disassembly procedure, or each procedure may be executed in a different order. Additionally,
the handling of sensitive or hazardous material (like mercury lamps) adds an extra level
of difficulty in the automation of the processing of e-waste. So far, humans seem to be