finite-horizon-lqr

To compute the first control input for finite-horizon LQR, provide your system matrices A, B, Q, R, the horizon N, and initial state x0. The solver performs backward Riccati recursion to calculate optimal feedback gains and returns the first input.

Yes, finite-horizon LQR is designed for model-predictive control workflows. It computes a short-horizon policy by calculating optimal state-feedback gains, making it suitable for real-time linear-quadratic control scenarios.

You need state matrix A, input matrix B, state cost matrix Q, input cost matrix R, the time horizon N, and the initial state x0 to run a finite-horizon LQR solver and generate the optimal control sequence.

Yes, finite-horizon LQR supports both linear time-invariant and time-varying systems. It applies backward Riccati recursion to compute the optimal sequence of state-feedback gains for your specified short horizon.

Yes, you need numpy installed to solve finite-horizon LQR problems using this method. The solver relies on numpy for handling the matrix operations required during backward Riccati recursion and forward simulation.

Finite-horizon LQR computes a distinct sequence of optimal state-feedback gains for each step over horizon N, whereas infinite-horizon LQR converges to a single steady-state gain. This makes finite-horizon suitable for short-horizon policies in MPC.