Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

Robot control software has rapidly evolved the past decade resulting in increasingly complex applications in increasingly dynamic environments. Nevertheless, most existing robot software architectures and frameworks are robocentric. If a resource has to be shared between multiple robots, no separate component or agent is defined to represent that resource. This lack of reflection of reality impedes robustness with respect to unexpected changes or disturbances and limits applicability.This dissertation presents a paradigm shift towards holonic robotic systems and explicit resource allocation, in which structural decomposition is essential. Consequently, physical entities, required in the application, receive their rightful place in the design, in contrast to the robocentric approach common in current robot software.In manufacturing control, holonic systems have earned their merit and resulted in more scalable and flexible systems, robust against unexpected disturbances A pioneer in this field of holonic systems is KU Leuven's Product-Resource-Order-Staff architecture (PROSA) and its short-term forecasting through delegate multi-agent systems (D-MAS). After being successfully applied to numerous manufacturing applications, this thesis investigates the applicability of PROSA and D-MAS to robotics.The scope of this research is limited to mobile service robotics and a multi-robot navigation scenario is presented to evaluate how well holonic systems fit in service robotics. A potential case study of this scenario is the deployment of a fleet of robotic wheelchairs in a hospital or retirement home, autonomously transporting patients from A to B.In the first part of the literature overview, concepts behind the holonics paradigm, in general, and PROSA and D-MAS, in specific, are introduced. Also, benefits of adopting structural decomposition and reflection of reality in the domain of robotics are identified.The second part reviews existing robot software architectures and frameworks, identifying their shortcomings with respect to separation of concerns and applicability. State of the art in multi-robot path planning and coordination is also evaluated keeping the proposed navigation scenario in mind.After this literature overview, design guidelines to implement holonic robotic applications are provided. The current PROSA implementation, the Holonic Manufacturing Execution System (HMES), and its integration to the existing mobile robot platforms at KU Leuven is also discussed. A set of use cases of holonic systems in service robotics is presented, out of which two are implemented.Multi-robot simulation experiments evaluate the complexity, scalability and robustness of the HMES. Scalability experiments demonstrate how the HMES copes with a larger set of navigation requests, on the one hand, and a larger environment, on the other hand. Finally, robustness against the unexpected is illustrated by blocking one of the corridors in the environment.Simulation experiments illustrate the benefits of resource awareness and making each shared resource in the environment a first-class citizen in the software design. Experimental data in simulation show that a coordinated approach with explicit resource allocation achieves better performance in path execution, measured by the defined benchmarks.The different experiments, in simulation and on the hardware platforms, clearly show the power of structural decomposition and short term forecasts that the HMES offers. The multi-robot path planning is distributed among these resources responsible for the planning, i.e. allocation, in the physical entity they represent. This largely simplifies the multi-robot path planning problem and increases scalability. The short term forecasts take care of unexpected disturbances to the resulting paths.This thesis provides a first step into applying holonic systems into service robotics and results in multi-robot navigation are promising. The road is paved to future research that further investigates the introduction of holons in other robotics domains such as manipulation or human robot interaction.

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

This thesis describes the design of an indoor visible light positioning (VLP) system. VLP is a technology to position a receiver (photodiode) based on the signals emitted by multiple transmitters (LEDs). This will be used in further research on VLP, more specifically for the localization of mobile robots. The system consists of a mechanical and an electrical part. For the mechanical part, a mounting base is designed. The purpose of the mounting base is not only to keep the LEDs in place, but also to make them moveable in three dimensions. Furthermore, they need to have the ability to tilt at certain angles. The position and pose measurement of the LEDs is included in the mechanical design. Multiple concepts are proposed for the design of the mounting base. These concepts are later ranked according to certain criteria. The electrical part consists of a drive circuit for the transmitters. The drive circuit will not only power the LEDs, but will also add the ability to modulate them. A possible method to use the setup for VLP is also introduced. The final goal of this thesis is to propose a final concept for a setup that can be used for VLP research at a later stage.

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)

This master thesis project deals with the development of control strategies for cooperatively transporting a flexible payload with multiple vehicles. To this end, an algorithm is developed which generates optimal trajectories for the vehicles to follow. The optimization problem solved by the algorithm to generate these trajectories is distributed, in the sense that each vehicle solves a copy of the problem to generate its own trajectory, while taking the trajectories of the other vehicles into account. The algorithm instructs that the optimization problem be solved repeatedly at different time instances while following the trajectories, making it fit into a distributed model predictive control (DMPC) framework. A flexible payload is designed and constructed, and 3 in-house holonomic vehicles are used for experimental validation of the algorithm.

Choose an application

• Reference Manager
• EndNote
• RefWorks (Direct export to RefWorks)