Modelling and optimization of the coordination of heterogeneous human-machine-teams

Project 2.2: Modelling and optimization of the coordination of heterogeneous human-machine-teams

Research Field 2: Simulation and optimization of heterogeneous human-machine-teams

PhD project 2.2: Modelling and optimization of the coordination of heterogeneous human-machine-teams in remanufacturing and recycling

Supervisors: Prof. S. Westphal, Prof. J. Zimmermann, and Prof. C. Schwindt

HMTs arise from the cooperation between humans and machines in the execution of tasks. In industrial production, human-machine-teams (HMTs) are particularly to be found where human intelligence and agility can be combined with machine power or precision. In contrast to traditional production systems, HMTs are characterized by greater flexibility, polyvalence, individuality, stochasticity and vulnerability. The coordination of HMTs includes the coordination between the machine and human elements of a team as well as the cooperation of several, generally heterogeneous teams. Depending on the chosen coordination approach, (partially) autonomous HMTs can pursue individual goals that contribute to the optimization of the overall system through appropriate incentive mechanisms. From the diversity of the teams, synergies can typically be achieved that do not occur when conventional systems are used.

An important future field for the use of HMTs is the remanufacturing and recycling of used products such as automotive modules, household appliances or hardware from ICT systems, which are processed into new products or from which recyclable components and secondary raw materials for the manufacture of new products are obtained. To this end, the waste equipment must be dismantled and its components tested and reconditioned with a view to their further use. A typical value-added chain after the collection of the old equipment consists of the stages of dismantling, reading, replacement or reprocessing of components and the recovery of secondary raw materials.

Comparatively high labor cost shares and health hazard potentials due to ergonomically demanding activities or the handling of hazardous substances predestine remanufacturing and recycling processes for automation as far as possible. Disassembly and reconditioning in particular, however, require a high degree of flexibility due to the large number of variants and the individual usage and wear characteristics of the devices and components. With the current state of automation technology, this flexibility cannot be fully or sufficiently reproduced economically, so that the use of human labor during dismantling cannot be dispensed with even in the medium term. Analogous to developments in assembly and intralogistics, HMTs are increasingly being used. Examples of this are the use of object recognition systems to support employees at identification workstations during reading, force assistance systems during intralogistics handling processes or collaborative robots during the dismantling of old equipment.

In the context of research into the modeling and optimization of heterogeneous HMTs, the scenario of a dismantling factory is to be investigated, which according to the principle of matrix production systems consists of a large number of individual dismantling cells flexibly connected in terms of material flow. The dismantling of an old device is typically carried out step-by-step via construction stages such as assemblies, components and individual parts. The execution of the respective dismantling step for all units of a dismantling order corresponds to one operation of the order. A dismantling cell has one or more collaborative robots, which, in cooperation with a worker, can disassemble different variants of devices or components, each in a single construction stage. A disassembly order therefore generally passes through several disassembly cells during its processing. Different technologies may be available for each operation, which differ, for example, in the degree of automation of the respective cells. The workers have polyvalent qualifications and can therefore be employed in different cells. However, they usually differ in terms of their skills profiles and work experience. Ideally, the corresponding value of a suitability parameter is known for each worker and each combination of dismantling cell and order type. Workers and the equipment of the dismantling cell form a temporary HMT for the duration of an operation.

The following decision problems occur during the control of the dismantling factory:

  • Selection of technologies for the operations of a dismantling order,
  • Allocation of suitable cells and workers to the operations of a dismantling order,
  • Sequencing of the dismantling orders in the worker-cell teams and scheduling of the operations.

The coordination of the execution of the dismantling orders by workers and cells can basically take place within the framework of central planning, rule-based decentralized control or self-control by the teams. Since dismantling orders are characterized by considerable uncertainties due to the difficulty in predicting the quality of the components, stochastic, robust, revising and online-planning methods can be used for central planning. A significant reduction in complexity results from the application of control systems that make allocation and sequencing decisions on the basis of simple priority rules or are based on the implementation of a control loop for production control according to the principle of Production Authorization Card systems. Most promising, however, is the coordination of HMTs by means of self-control, in which the workers take on the role of (partially) autonomous agents competing for the combinations of operations, cells and sequence positions in which they can process the operations assigned to them in the respective cells. For dismantling orders with several operations, the cooperation of several teams will generally make sense in order to exploit synergies. Such self-control can be achieved, for example, via the mechanism of combinatorial auctions in which workers or groups of workers submit bids for combinations of dismantling orders, cells and sequence items. The allocation of the offers to the bids can be made in such a way that a suitable overarching objective such as the maximization of the sum of the realized suitability characteristic values is pursued. The allocation mechanism must be combined with a scheme for the distribution of profits among workers, which offers suitable incentives for cooperative behavior and enables as fair a distribution as possible. In addition to operational control of the dismantling factory, self-control is also suitable for increasing employee motivation through the transfer of decision-making authority and for providing long-term incentives for independent further qualification.

The object of the investigations will be the development and simulation-supported comparison of suitable methods for the coordination of heterogeneous HMTs based on the paradigms of central planning, decentralized control and self-control of the teams using the example of a dismantling factory. For different application scenarios, the strengths and weaknesses of the respective coordination forms are to be identified and recommendations for their use are to be developed.