What to expect

A Multi-Agent System is a network of physically uncoupled subsystems that are connected via a communication structure. To ensure the scalability and availability of such systems, the agents are controlled decentrally by networked controllers. Equipping mobile robots and UAV with gas sensors enables detection, localization and tracking of the releases of gasses into the atmosphere and of the resulting gas plumes. The control of such a multi-agent gas detection system needs to take into account constraints both from the coupling structure, as well as the limitations of the sensing elements. Simultaneity and spacial distribution of sensing nodes can offer a way to tackle the highly dynamic nature of airborne gas dispersion processes.

Your tasks

Your thesis with the DLR Swarm Exploration Group will revolve around distributed formation control for a multi-agent gas detection system. Your work will focus on the development of a control scheme that solves the combined constraints of distributed control, limited sensor fidelity and highly dynamic gas dispersion. It will include simulation of the multi-agent system and your control strategies, and evaluation of several candidate strategies that leverage certain flight formations to gain information on atmospheric gas releases. While primarily simulation-focused, this thesis can optionally be extended to entail evaluation of your algorithms in field experiments with our drones and a synthetic gas source.

Your profile

• Studies in electrical engineering, information technology or in related fields
• good knowledge of control theory, preferably with experience in multi-agent control
• excellent programming skills in Python
• knowledge of gas dispersion physics, gas sensor technology are a plus but not mandatory
• self-motivated working style and a good knowledge of English or German