Point-to-point control for nonlinear systems (Q597268)

The paper deals with the point-to-point control problem of nonlinear systems. Consider the nonlinear system given by \[ \frac{dx_i(t)}{dt}=f_i(x(t),v(t)),\quad t>0,\;i=1,2,\dots,n,\;x(0)=x_0\tag{1} \] where \(x(t)\in\mathbb R^n\), \(v(t)\in\mathbb R^m\), \(t\geq0\), \(n\geq m\) and \(f_i(x,v)\), \(i=1,2,\dots,n\), are smooth functions on \(\mathbb R^n\times\mathbb R^m\). Given a vector \(x_D\in\mathbb R^n\), \(x_D \neq x_0\) and the time \(t_f\), \(0<t_f <\infty\). The point-to-point control problem is to find a control function \(v(.)\), if it exists, such that when \(v(.)\) is acting as an input to the system given above, the condition \(x(t_f) = x_D\) will be satisfied. It is shown, by using optimal control and by solving numerically an optimization problem, that the point-to-point problem has a solution for two controllable nonlinear systems. However, the numerical study indicates that for each two states \(x_0\) and \(x_D\) there exists only a range of terminal times \(t_f\) for which the problem can be solved. The controllability of the nonlinear system for a special form of (1) is investigated. Two examples (the motion of a cart and the motion of a trolley) are numerically investigated in detail.