Planning the Actions of a CoBOT adapted to human variabilities – pacbot

An important challenge for the factory of the future is to preserve the health and comfort of operators while improving their productivity. At the heart of this challenge, people expertise guarantees the flexibility and know-how of companies. Collaborative robotics or cobotics is a well-identified solution that should allow, through appropriate assistance, to focus the operator on the tasks of which he/she has the expertise, while delegating loads and constraints to an intelligent tool. The strong idea of the Pacbot project is that the operator must keep a great freedom of movement (preserving the movement variability and strategies of the human) and of decision (preserving the variability in the order of the subtasks to perform a complex task). These points are important to guarantee the acceptance of the cobot. The ability of the cobot to adapt in terms of coordinated cognitive assistance (orchestration and distribution of tasks and subtasks, autonomy and shared situational awareness) and inferred physical assistance (movements, gestures, postures) is a sine qua non condition for its acceptance. In particular, it must improve: 1) the comfort of the operator; 2) the natural communication between the human and the cobotic assistant; 3) work in symbiosis; 4) the expertise of the operator, at least to preserve it; 5) the quality and productivity of shared work while alerting the operator of physical or cognitive load with the purpose of preventing risks for his/her health.

In this context, the general objective of the Pacbot project is to design a semi-autonomous cobotic system for operator assistance in industrial settings, capable of selecting, synchronizing and coordinating tasks distributed between human and cobot by adapting to different types of variability in professional gestures, while anticipating dangerous situations. The objective here is to limit human errors related to poor situational awareness by adapting the behavior of the cobot to the operator and not the other way around. Concretely, the scientific objective of the Pacbot project is twofold:

1- devise algorithms capable of constructing plans (sequences of decisions and movements) taking into account the intra-personal variability of the operators. AI Planning will be used as a decision engine both to predict the operator's behaviour and to decide autonomously on the actions that the cobot will have to perform. We will integrate in the task and motion planning process a risk criterion associated with a potential reduction of situation awareness.

2- identify sources of variability favourable to health and work performance, perform the analysis of the social acceptance of the cobot and develop a design method that fully integrates the user experience from the technical design of a new cobotic system. The production of methodological tools for ergonomists and occupational psychologists is an important expectation of the project: to provide an original method of designing workstations for the factories of the future with the introduction of a cobot and the co-construction of programming with computer scientists.

The Pacbot project is an original interdisciplinary project based on the strong and balanced involvement of Social Sciences and Humanities researchers (ergonomists and occupational psychologists) and computer scientists. The relevance of the methodological and technical proposals produced during the project will be demonstrated through the realization of a prototype of a cobot workstation integrating the determinants of intra-personal variability in a man-robot collaboration situation within a glove box for handling chemicals in partnership with the Solvay company.