We discuss the problem of local attainability for finite-dimensional nonlinear control systems with quite general assumptions on the target set. Special emphasis is given to control-affine systems with a possibly nontrivial drift term. To this end, we provide some sufficient conditions ensuring local attainability, which involve geometric properties both of the target itself (such as a notion of generalized curvature), and of the Lie algebra associated with the control system. The main technique used is a convenient representation formula for the power expansion of the distance function along the trajectories, made at points sufficiently near to the target set.We notice that Petrov's condition is a first order condition in the sense that it involves only admissible velocities.From another point of view, this condition requires d S to be a sort of Lyapunov function for the system. Indeed, from a geometrical point of view, Petrov's condition states that at every point x of a neighborhood of the target there exists an admissible control u x ∈ U such that the corresponding trajectory points sufficiently toward the target. Moreover, the scalar product between the admissible velocity f (x, u x ) and the gradient of the distance is uniformly bounded away from zero.Petrov's condition is very strong, even if it is weaker than full controllability, which requires that every initial state can be steered to any final state in finite time along admissible trajectories. Moreover, it can be also shown that Petrov's condition is equivalent to the Lipschitz continuity of the minimal time function T up to the boundary of S (see [23]). However, it is also very easy to give simple examples where it fails. For instance, in R 2 take S = {0} and (ẋ(t),ẏ(t)) = (y(t), u(t)), where u : R → [−1, 1] is measurable: Petrov's condition fails on the x-axis.1.3. Higher order condition for pointwise target. When Petrov's condition is not satisfied, i.e., the trajectories of the system do not approach the target at the first order, it is natural to search for higher order conditions, which involve higher order terms in a convenient expansion of the trajectory itself. These conditions will be related to some properties of the Lie algebra generated by the family of vector fields associated with the system (see [15] for a complete introduction).In the early 1960s, Kalman proved the following result. Assume that f is linear, i.e., f (x, u) = Ax + Bu, where A ∈ Mat n×n (R), B ∈ Mat n×m (R) are two constant matrices, and S = {0}. Then the following are equivalent: 1. the system is controllable to the equilibrium point 0, i.e., every point can be steered to the origin in finite time; 2. the matrix (B|AB|A 2 B| . . . |A n−1 B) has full rank (equals n). The second condition above is the celebrated Kalman rank condition, and implies the Hölder continuity of T , with exponent depending on the smallest 0 ≤ k ≤ n − 1 such that the matrixhas full rank.Later, in the 1970s, several generalizations, mainly concerning the case when target set S is an equilibrium po...