COOPERATIVE IMPLICATIONS OF THE ALLOCATION OF FUNCTIONS TO HUMANS AND MACHINES

Objective: This paper aims at pointing out the cooperative aspects of function allocation to humans and machines in complex systems. Background: It is based on the observation that interference exists mostly between functions devoted to each of these agents. Furthermore, when interfering functions are distributed among several agents, the latter need to perform cooperative activities to manage the former. Method: It uses functional, structural, and social frameworks to distinguish between various levels and forms of cooperation, and different attitudes toward automation. Such frameworks allow us to consider the cooperative implications of the main FA approaches: Fitt’s MABA-MABA method, the LOA (Level Of Automation) taxonomy and the DFA (Dynamic Function Allocation) approach. Results: It emphasizes the adaptive ability of humans who are given supervision of the HMS, and consider FA in terms of function delegation. This paper emphasizes the need to elaborate and maintain a COFOR (COmmon Frame Of Reference) to manage interference. It puts forward the cost of cooperative activity and the need to evaluate its efficiency in terms of the relationship between cost and benefits. Conclusion: Finally, it proposes the MOC (Mode Of Cooperation) framework as an alternative to LOA and presents an application of MOC to cardriving assistance domain. Application: The paper identifies strategic decisions in the design of Human-Machine Systems in order to improve Human-Machine Cooperation and system safety.