Implementation of iLQR (Iterative Linear Quadratic Regulator) algorithm for trajectory optimization, based on Yuval Tassa's Matlab implementation.
See notes.md for comments on algorithm, implementation, to-do's, etc.
Mostly working, except the "control-limited" part... See notes for details. If you're looking for a C++ implementation with control limits, the best one I've found is Ben Stephen's implementation (I haven't tried it though).
- Install eigen or download into
include(https://eigen.tuxfamily.org/index.php?title=Main_Page#Download) mkdir build; cd buildcmake ..make- Define a dynamics and cost model based on Model (see double_integrator example), or run with
./run_iLQR acrobot
- Tassa, Yuval, Nicolas Mansard, and Emo Todorov. "Control-limited differential dynamic programming." Robotics and Automation (ICRA), 2014 IEEE International Conference on. IEEE, 2014.
- Li, Weiwei, and Emanuel Todorov. "Iterative linear quadratic regulator design for nonlinear biological movement systems." ICINCO (1). 2004.