< Back to previous page

Project

Situational aware autonomous systems with online task coordination and negotiation

This PhD project will be part of a larger Flanders Make SBO project, AssemblyRecon, that tackles current issues with reconfigurable assembly system (RAS) configurations. They typically result in a suboptimal space and capital expenditure utilization as (1) they are not continuously operating at full potential (e.g. volume changes due to product ramp-up, market fluctuations or end-of-life) and (2) they need to be re-engineered drastically and frequently (e.g. new products or volume changes). AssemblyRecon will focus on Reconfigurable Assembly Systems and tackle the following barriers, which hamper their industrial adoption: - Lack of a decision framework to decide on assembly plant reconfiguration at three levels: workstation level (1 day to 1 month), system architecture management level (1 day to 1 month) and on-line task execution level (1 to 10 seconds). The former two respond to production changes, such as product mix, variants or volume, while the latter responds to incoming assembly orders and possible execution disturbances, such as stock shortage, rush orders, quality issues and breakdowns. - Lack of a proof of concept in a relevant environment showcasing the potential of RAS in terms of optimization of capex & space utilization enabled by a novel decision framework and well-established technologies (regarding flexible flow systems, inline kitting and picking systems, industrial machine to machine communication protocols and new design concepts for flexible workstations) In order to do so, the project will deliver both an Assembly Configuration Recommender (ACR) and an Assembly Execution System (AES), which will closely interact. The ACR will propose reconfiguration based on external triggers extracted from production changes using a two-step approach. In a first step initial configurations will be proposed based on historic data and configurations. To this end, a graph-based database approach will be realized, containing a.o. models describing the assembly system characteristics and capabilities. The ACR can call upon six different optimization modules in two categories (workstation and system architecture) to compute optimal assembly system configurations, applying (meta-) heuristic approaches smartly combined with generic local search engines. The AES will follow a distributed approach in which incoming assembly orders are translated in dynamic task execution commands for the respective hardware modules. Continuously, each module will indicate its availability and status. As each module is owner of its own time slots, fast local optimizations are possible in case of disturbances, such as breakdowns or stock shortages. The AES will continuously update stochastic models to estimate performance, such as task execution times, reduce uncertainty and predict the need for global reconfiguration requests towards the ACR. This PhD project will focus on the Assembly Execution System (AES) and, thus, the selected candidate will also have to closely work together with researchers developing the ACR. Important research targets in this PhD project are (1) semantic modelling of task execution in an assembly context; (2) resource allocation (scheduling) based on the higher-level decisions (selected layout design, workstation and transportation options); (3) coping with external (e.g. stock shortage, rush orders, …) and internal (e.g. hardware failures, assembly failures, delays, …) disturbances during assembly execution, and design probabilistic models for performance estimation and prediction to improve the robustness of the AES to such disturbances.

Date:1 Oct 2020 →  Today
Keywords:Robotics, Coordination, Modelling, Resource allocation
Disciplines:Adaptive agents and intelligent robotics, Mobile and distributed robotics
Project type:PhD project