Mathematical Decision Making (MDM)

In the group of Mathematical Decision Making (MDM), we develop novel control-theoretic methods, especially based on model predictive control (MPC), for real-time autonomous control of systems. These developments emphasize effective, computationally-efficient handling of uncertainties in unknown and dangerous environments.

A core application that MDM focuses on is mission-planning for search-and-rescue robots, for various tasks including data fusion and mapping of unknown environments, finding and tracking trapped victims, and coordinating multiple (non-homogeneous) robots to minimize the time of search-and-rescue and maximize the number of the detected victims and the coverage of the area.

We also implement and assess these approaches via real robots.

MDM projects

• State-dependent dynamic tube model predictive control (SDD-TMPC)

  A new dynamic version of robust tube MPC that incorporates the dynamics of disturbances into decision-making.

• Flying and Ground Robot Collaboration for Camera-based Search and Rescue

  A cost-effective framework combining UAV and UGV video streams for search-and-rescue tasks.

• Non-homogeneous autonomous robots for search and rescue in the aftermath of disasters

  Novel mathematical reformulations of the control optimization problems of search-and-rescue robots.

• Bi-level temporally-distributed MPC

  A bi-level temporal distribution of complex MPC optimization problems for real-time control of urban traffic networks.

• Tube MPC with dynamic obstacles

  Tube model predictive control for motion planning of robots in environments with dynamic obstacles.