real-time performance
real-time performance
Embedded Files
To demonstrate the aptitude of ILQ-games to solve real problems in real-time, we set up a 3D simulation for the control of multi-agent systems over a receding time-horizon.
Developed using Robot Operating System (ROS) with RVIZ for well documented, easy to implement 3D simulations.
Written in C++ to insure optimal performance.
Three Modules
Three Modules
The Planner module accepts the state of each player at a given time in the simulation and calculates the best trajectories and strategies for the agents.
The controller module takes strategies from the planner and converts them into control inputs for the simulator
The Simulator takes control inputs and propagates each players dynamics forward.
The Simulation
The Simulation
This ongoing work, though not yet fully implemented for a general case, is promising, and clearly demonstrates the solvers ability to consistently solve complicated problems in as little as 100 ms, sufficient to be implemented in real-time for tough multi-agent problems.
Page updated
Google Sites
Report abuse