1
$\begingroup$

Bézier curves have some nice properties, such as starting at $P_0$ and ending at $P_n$ (for an $n$-degree Bézier curve). I am looking for a class of (curvy) curves, but with the additional property that their arc-length is always equal to a given $L$ with $L>L_\mathrm{min}$, where $L_\mathrm{min}$ is the Euclidian distance between $P_0$ and $P_n$.

I have tried to achieve this by changing the weights of intermediate points, this works but it is more of a numerical approach (and is computationally intensive).

Is there a class of curves that allows me to prescribe their arc-length? While still being able to generate distinct and curvy curves?

$\endgroup$
3
  • 1
    $\begingroup$ I would say no 1) by my experience in this domain, I haven't met such curves 2) because arclength involves rather complicated formulations. But a rather satisfying solution would be to consider curves made of (an arbitrary number of) connected circular arcs with smooth connections ; a pitfall is that you loose the differentiability at any order... $\endgroup$ Commented Jan 2, 2022 at 11:41
  • $\begingroup$ @JeanMarie thank you, this is incredibly helpful. Is there by any chance a specific name for this method using circular arcs? Dubin's paths seem related $\endgroup$ Commented Jan 2, 2022 at 13:17
  • 1
    $\begingroup$ This isn't what you're asking for, but you may like to investigate the Cornu / Euler spiral, "a curve whose curvature changes linearly with its curve length". Its properties make it essential for applications like roads, railway tracks, and roller-coasters. $\endgroup$ Commented Jan 2, 2022 at 16:35

2 Answers 2

Reset to default
2
$\begingroup$

An alternative family of curves having easily computable arc lengths are curves made of smoothly connected circular arcs.

One must know that that two circular arcs with centers in $C_1$ and $C_2$ are smoothly connected at point $P$ iff points $P,C_1,C_2$ are aligned.

I have been working on this type of curves. Here are three examples of them in pictures:

enter image description here

Fig. 1: Wheels. The LHS "wheel" is build out a general triangle with sidelengths $a,b,c$. Sidelength $a$ is extended by 2 line segments: one having length $c$, the other one having length $b$, in order that the total length of the extended side is $a+b+c$. This operation is done twice again. Whatever the direction, the apparent diameter is always $a+b+c$. The RHS figure, instead of a triangle, is based on a (starshaped) pentagon. See here.

enter image description here

Fig. 2: Two different methods for the drawing of ovals taken from the reference book of Serlio (Italy, around 1550). In some particular cases, they approximate in a correct way ellipses. Please note that they are based on "pure" figures (squares and equilateral triangles) in the euclidean tradition. The third figure shows that a supplementary degree of freedom is possible by dilating previously obtained figures, just as what we have done in fig. 1 for the "wheels".

References for ellipse and spiral drawings with circular arcs technique: http://evlm.stuba.sk/APLIMAT2018/proceedings/Papers/0450_Holesova.pdf http://users.cs.cf.ac.uk/Paul.Rosin/resources/papers/oval2.pdf

http://faculty.evansville.edu/ck6/ellipse.pdf

What kind of mathematical spiral does the volute of an Ionic capital approximate?

enter image description here

Fig. 3: "Avoiding curves" for two aircrafts in red and green arriving from NW and NE resp. with a colliding risk if their trajectories aren't changed. The air control regulation will schedule for them a "roundabout routing" in order to avoid this collision risk. Here the "green" aircraft will take a longer path in order to let the "red" aircraft pass before it.

Remark: I didn't knew Dubins paths. Thank you.

$\endgroup$
3
  • 1
    $\begingroup$ I see a red and green path, not blue and green. Maybe there should be a blue one? $\endgroup$ Commented Jan 2, 2022 at 16:15
  • $\begingroup$ @soupless Sorry. Fixed now. $\endgroup$ Commented Jan 2, 2022 at 20:49
  • $\begingroup$ I have improved my presentation. Btw, I am not aware of any specific name for this kind of curves. $\endgroup$ Commented Jan 3, 2022 at 9:20
1
$\begingroup$

You can make smoothly connected strings of circular arcs using the so-called biarc technique.

You can find quite a bit of information by searching for that term on this site and elsewhere. For example this question.

The answers include several useful links.

One of the links provides a very detailed explanation, plus C++ code for biarc approximation.

Another option is a special type of Bézier curve called a Pythagorean hodograph (PH) curve. These curves have the nice property that their arclengths can be computed quickly and easily by simple formulas (whereas general Bézier curves require numerical integration techniques). A PH curve of degree 5 has roughly the same controls as a regular cubic Bézier curve — you can specify its initial and final positions and tangents. There’s an entire book about PH curves written by Rida Farouki.

$\endgroup$

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.