Mixed-Integer Linear Programming for Control

Department of Aeronautics and Astronautics
Aerospace Controls Laboratory
Massachusetts Institute of Technology

Mixed-integer linear programming (MILP) is a very general framework for capturing problems with both discrete decisions and continuous variables. This includes:

• Assignment problems
• Control of hybrid systems
• Piecewise-affine (PWA) systems (including approximations of nonlinear systems)
• Problems with non-convex constraints (e.g., collision avoidance)

You can obtain some sample code for trajectory design and assignment problem for UAV guidance. It requires Matlab and AMPL/CPLEX system to run the code. The MILP problems are sized down so that AMPL/CPLEX student version can solve them.

• Trajectory optimization for an autonomous UAV
  Developed a new approach to trajectory optimization for autonomous aerial vehicles.
  The planner can design a trajectory that reaches targets, while satisfying constraints such as no-fly zones and aircraft dynamics.
  
  • Sample code
  • Sample results ( 1 , 2 , 3 )
  
  
• Task assignment for a fleet of UAVs
  Develop a general methodology for the team tasking and cooperative path planning
  components of hierarchical coordination architecture for teams of UAV's.
  
  • Sample code
  • Sample result
  
  
• Recent Conference Papers by the research group.