“…Although Markov-Dubins path has been extensively studied both theoretically and practically, finding as wide range of applications as the path planning of drones (or uninhabited aerial vehicles) and robots, and the tunnelling in underground mines, to the author's knowledge, its generalization to interpolation has not been studied in its entirety or true form yet. Brunnett, Kiefer and Wendt [8] consider Problem (P) (but not in the form we pose it here) with a bound on the average, rather than the pointwise, curvature, and propose an algorithm for finding what they call a "Dubins spline" with the following ad hoc steps: (Step 1) Estimate/guess the missing velocity directions at the interior nodes, (Step 2) Find a Markov-Dubins path of type CSC (but not of type CCC or a subset thereof) between each two consecutive nodes and (Step 3) Update the estimated missing velocity directions at the interior nodes using some nonlinear optimization procedure to minimize the overall length of the spline; [if some stopping criterion is not satisfied, then] go to Step 2. In [8], not only the problem that is solved is different from Problem (P), but also the theory and pertaining analysis are not adequately covered. Similar ad hoc approaches ultimately leading to interpolants which are suboptimal solutions, of problems related to Problem (P), can be found in the literature; see, for example, [20,29,37].…”