Sono state proposte numerose tassonomie e sistemi di classificazione dei livelli di automazione (LOA). Lo schema di classificazione più antico è quello propo- sto da Sheridan e Verplank alla fine degli anni settanta88. Esso prevede una scala
di dieci punti, che rappresentano livelli crescenti di automazione. Al livello più basso della scala, l’essere umano prende decisioni e agisce senza alcuna assi- stenza. Al livello più alto, la macchina prende decisioni e agisce in completa autonomia, come mostrato nella tabella2.1.
Tuttavia, la scala usata in questa prima tassonomia si limita a determinare una serie di punti distinguibili, lungo livelli continui di automazione, principal- mente in relazione alle funzioni di output del processo decisionale e della scelta
85In questo senso Hollnagel, E. «From function allocation to function congruence». In: Coping
with computers in the cockpit(A 00-40958 11-54), Aldershot, United Kingdom and Brookfield, VT, Ashgate Publishing, 1999, (1999), pp. 29–53.
86Fitts, P. M. «Human engineering for an effective air-navigation and traffic-control system.»
In: (1951). Questi elenchi si basano sull’idea che, dato un insieme di attività preesistenti si decide quali di queste rendere automatiche, considerando i punti di forza di esseri umani e macchine
87Endsley,«Automation and situation awareness», cit.
TABELLA 2.1: Levels of Automation of decision and action
selection - Sheridan and Verplanck (1978)
Automation Level Automation Description
1 The computer offers no assistence: human must take all decisions and actions 2 The computer offers a complete set of decisions and actions alternatives, or 3 narrows the selection down to a few, or
4 suggest one alternative and
5 executes that suggestion if the human approves, or
6 allows the human a restricted time to veto before automatic execution, or 7 executes automatically, then necessarily informs human and
8 informs the human only if asked, or
9 informs the human only if it, the computer decides to.
10 The computer decides everything and acts autonomously, ignoring the human
del tipo di azione da mettere in atto. Mancava una specificazione dettagliata delle funzioni di input, relativa all’acquisizione delle informazioni.
Negli stessi anni Endsley e Kaber89costruiscono una tassonomia su dieci li-
velli, applicabile a una vasta gamma di settori e tipi di attività, riportata nella tabella2.2. La loro tassonomia comprende quattro funzioni generiche, compa- rabili a quelle successivamente elaborate da Parasuraman, Sheridan, e Wickens, dove a ogni livello corrisponde una funzione o una combinazione di funzioni sia dell’essere umano, sia del sistema automatico.
TABELLA2.2: Endsley and Kaber’s (1999) - LOA taxonomy for
human–computer performance in dynamic, multitask scenarios.
Roles
Level of Automation Monitoring Generating Selecting Implementing
(1) Manual Control Human Human Human Human
(2) Action support Human/Computer Human Human Human/Computer
(3) Batch processing Human/Computer Human Human Computer
(4) Shared control Human/Computer Human/Computer Human Human/Computer
(5) Decision support Human/Computer Human/Computer Human Computer
(6) Blended decision making Human/Computer Human/Computer Human/Computer Computer
(7) Rigid system Human/Computer Computer Human Computer
(8) Automated decision making Human/Computer Human/Computer Computer Computer
(9) Supervisory control Human/Computer Computer Computer Computer
(10) Full automation Human Computer Computer Computer
89Endsley, M. R. «Level of automation effects on performance, situation awareness and
Un passo decisivo è stato fatto da Parasuraman, Sheridan, e Wickens90, che,
sulla base della scala elaborata da Sheridan e Verplank, hanno introdotto l’idea di associare i livelli di automazione ai tipi di funzione, come riportato nella figu- ra2.2. Le funzioni sono modellate sulla base di quattro fasi distinte, necessarie affinché un essere umano sia in grado di processare informazioni e agire: (1) acquisizione dei dati, (2) analisi delle informazioni, (3) decisione e scelta dell’a- zione, e infine (4) attuazione. Le quattro fasi possono essere tradotte in funzioni equivalenti del sistema e ogni funzione può essere automatizzata secondo livelli differenti di automazione. Automation Level Automation Level Automation Level Automation Level
High High High High
Low Low Low Low
Information Acquisition Information Analysis Decision Selection Action Implementation System B System A
FIGURA 2.2: A model for Types and Levels of Automation
proposed by Parasuraman, Sheridan and Wickens (2000)
Più di recente, e all’interno di un progetto SESAR (Single European Sky ATM
90Parasuraman, R., Sheridan, T. B. e Wickens, C. D. «A model for types and levels of human
interaction with automation». In: Systems, Man and Cybernetics, Part A: Systems and Humans, IEEE Transactions on 30.3 (2000), pp. 286–297.
Research)91 nel settore aeronautico, chiamato "Good Practices for HP Automa-
tion Support", e sulla base dei modelli precedenti, Save e Feuerberg 92 hanno
presentato una nuova tassonomia dei livelli di automazione. Per ogni funzione sono stati sviluppati differenti livelli di automazione. La tassonomia -riportata nella tabella 2.3- si presenta suddivisa in quattro colonne, corrispondenti al- le quattro funzioni generiche elaborate da Parasuraman, Sheridan, e Wickens. Ogni funzione presenta un diverso numero di livelli di automazione: cinque per l’acquisizione dei dati e l’analisi delle informazioni, sei per la decisione e scelta dell’azione e infine otto per l’attuazione.
TABELLA2.3: The Level of Automation Taxonomy (LOAT)-Save
e Feuerberg (2012)
A B C D
Information Acqui-
sition Information Analy-sis Decision and ActionSelection Action Implementa-tion
A0 Manual Informa-
tion Acquisition B0 Working Memo-ry Based Informa-
tion Analysis
C0 Human Decision
Making D0 Manual Actionand Control
The human acquires relevant information on the process s/he is following without using any tool.
The human compa- res, combines, and analyses different in- formation items re- garding the status of the process s/he is following by way of mental elaborations. S/he does not use any tool or support external to her/his working memory.
The human generates decision options, se- lects the appropriate ones and decides all actions to be perfor- med.
The human executes and controls all ac- tions manually. A1 Artefact- Supported information Acquisition B1 Artefact- Supported Information Analysis C1 Artefact- Supported Decision Making D1 Artefact- Supported Action Implementation
91SESAR è un progetto a cui collaborano la Commissione Europea ed Eurocontrol, insieme ad
oltre cento partner del settore. L’obiettivo è modernizzare l’ATM, il sistema di gestione del traffico aereo.
92Save, L., Feuerberg, B. e Avia, E. «Designing human-automation interaction: a new level of
The human acquires relevant information on the process s/he is following with the support of low-tech non-digital artefacts.
The human compa- res, combines, and analyses different information items regarding the status of the process s/he is following utili- sing paper or other non-digital artefacts.
The human generates decision options, se- lects the appropriate ones and decides all actions to be perfor- med utilising paper or other non-digital artefacts.
The human exe- cutes and controls actions with the help of mechanical non-software based tools.
A2 Low-Level Auto- mation Support of Information Acqui- sition B2 Low-Level Auto- mation Support of Information Analy- sis C2 Automated Deci-
sion Support D2 Step-by-step Ac-tion Support:
The system supports the human in acqui- ring information on the process s/he is following. Filtering and/or highlighting of the most relevant information are up to the human.
Based on user’s request, the system helps the human in comparing, combi- ning and analysing different information items regarding the status of the process being followed.
The system proposes one or more decision alternatives to the human, leaving free- dom to the human to generate alternative options. The human can select one of the alternatives propo- sed by the system or her/his own one. The system assists the operator in per- forming actions by executing part of the action and/or by providing guidance for its execution. However, each action is executed based on human initiative and the human keeps full control of its execution.
A3 Medium-Level Automation Sup- port of Information Acquisition B3 Medium-Level Automation Sup- port of Information Analysis C3 Rigid Automated
Decision Support D3 Low-Level Sup-port of Action Se-
quence Execution
The system supports the human in acqui- ring information on the process s/he is following. It helps the human in inte- grating data coming from different sour- ces and in filtering and/or highlighting the most relevant in- formation items, ba- sed on user’s set- tings.
Based on user’s request, the system helps the human in comparing, combi- ning and analysing different information items regarding the status of the process being followed. The system triggers vi- sual and/or aural alerts if the analysis produces results requiring attention by the user.
The system propo- ses one or more de- cision alternatives to the human. The hu- man can only select one of the alternati- ves or ask the sy- stem to generate new options.
The system performs automatically a se- quence of actions af- ter activation by the human. The human maintains full con- trol of the sequence and can modify or in- terrupt the sequence during its execution.
A4 High-Level Au- tomation Support of Information Acqui- sition B4 High-Level Au- tomation Support of Information Analy- sis C4 Low-Level Auto- matic Decision Ma- king
D4 High-Level Sup- port of Action Se- quence Execution
The system supports the human in ac- quiring information on the process s/he is following. The system integrates data coming from different sources and filters and/or highlights the infor- mation items which are considered re- levant for the user. The criteria for inte- grating, filtering and highlighting the rele- vant information are predefined at design level but visible to the user.
The system helps the human in comparing, combining and ana- lysing different infor- mation items regar- ding the status of the process being follo- wed, based on pa- rameters pre-defined by the user. The sy- stem triggers visual and/or aural alerts if the analysis produ- ces results requiring attention by the user.
The system genera- tes options and deci- des autonomously on the actions to be per- formed. The hu- man is informed of its decision.
The system performs automatically a se- quence of actions af- ter activation by the human. The human can monitor all the sequence and can in- terrupt it during its execution.
A5 Full Automation Support of Informa- tion Acquisition B5 Full Automation Support of Informa- tion Analysis C5 High-Level Auto- matic Decision Ma- king
D5 Low-Level Au- tomation of Action Sequence Execution
The system supports the human in ac- quiring information on the process s/he is following. The system integrates data coming from different sources and filters and/or highlights the infor- mation items which are considered re- levant for the user. The criteria for in- tegrating, filtering and highlighting the relevant info are predefined at design level and not visible to the user (transpa- rent to the user in Computer Science terms).
The system performs comparisons and analyses of data avai- lable on the status of the process being followed based on parameters defined at design level. The system triggers vi- sual and/or aural alerts if the analysis produces results requiring attention by the user.
The system genera- tes options and deci- des autonomously on the action to be per- formed. The hu- man is informed of its decision only on request.
The system initiates and executes auto- matically a sequence of actions. The hu- man can monitor all the sequence and can modify or interrupt it during its execution.
C6 Full Automatic
Decision Making D6AutomationMedium-Levelof
Action Sequence
Execution
The system genera- tes options and deci- des autonomously on the action to be per- formed without in- forming the human. (Note that this le- vel is always con- nected to some kind of ACTION IMPLE- MENTATION, at an automation level not lower than D5.)
The system initiates and executes auto- matically a sequence of actions. The hu- man can monitor all the sequence and can interrupt it during its execution.
D7 High-Level Au- tomation of Action Sequence Execution
l|The system initia- tes and executes a sequence of actions. The human can only monitor part of it and has limited opportu- nities to interrupt it.
D8 Full Automation of Action Sequence Execution
l|The system initia- tes and executes a sequence of actions. The human cannot monitor nor interrupt it until the sequence is not terminated.
Questa tassonomia sarà utilizzata nell’ultimo capitolo per l’analisi di uno scenario, definendo in che modo attività e compiti sono distribuiti tra uomo e macchina e analizzando le possibili responsabilità degli attori coinvolti.