Section: New Results

New method for cobotic task evaluation

Two industrial studies allowed us improving our methodology for cobotic task evaluation.

• Thanks to the partnership with Suez and the work of ENSC student Nina Docteur under Auctus supervision, there are several interesting results: first, the methodological approach has been reinforced. There was a detailed analysis of an accident-prone gesture (the pipe cover raising), meetings with field agents, supervisory teams, discussions with SUEZ ergonomic expert and field observations. Second, there was a theoretical framework - a model - for the general evaluation of a cobotic task, as well as the exploration of rules for evaluating the components of this model. Five main components have been proposed for the evaluation: bio-mechanics, cognitics, usability, hedonism and social impact. An important difficulty was to mix every component and to unify the evaluation. In order to mature the model, it has been decided to carry on the partnership with a PhD work.

• The PORTAGE project (Plateforme de RoboTisation et d’Automatisation GénériquE de bâtis industriels) involves AKKA Technologies, Ez-Wheel, IIDRE and IMS laboratory. It aims at developing semi-autonomous solutions for moving heavy structures within industrial environments (e.g. aircraft industry). Our contributions is concerned with human-robot interactions, and especially accounting for real-life constraints of operators’ job within their industrial environment, and translate them into design choices and requirements for the to-be-developed robotic solution. In order to identify relevant elements from the work situation, three Human Factors models have been used: Reason’s model [54], Situation Awareness model [27], and Skill Rule Knowledge (SRK) model [53]. The Reason’s model details the different layers to explain accidents, notably in the aircraft industry. These layers gather equipment, procedure, training, management, policies and even psychological precursors of the operator. Therefore, this model allows investigations on potential latent causes of accident in complement with “obvious” patent causes usually more easily identified. The Situation Awareness model of Endsley describes cognitive mechanisms involved in a given situation for a person when performing actions, based on 1) the perception of elements of the current situation, 2) the understanding of the current situation, and 3) the projection of the future status of the situation. This model leads to identifying 8 daemons where situation awareness can be deteriorated, potentially resulting in accidents. The SRK model describes the decision process of an operator (or any person) performing a given task, based on his/her familiarity with this task. This model, coupled with the Situation Awareness model, can be leveraged to identify elements to be accounted when developing collaborative robots in industrial environments.