Reaching target as fast as possible

What you know:

  • Max acceleration/deceleration (can assume them to be equivalent
  • Max speed
  • Distance to travel

You want to get to the end as fast as possible and end with 0 velocity
Also I want it to look natural, not just an abrupt stop by losing max acceleration every time (this is the part I’m having trouble with)

How do you do this?

I feel like you’re supposed to use a spring for this but the issue is that it has to be based on speed and position (?) because you want it to maintain the max speed but you also don’t want to go past the target so maybe you have to use 2? lol


Is the rate of change in acceleration linear?

It can be but I want it to look natural so I’m not sure if that would be ideal

This is probably equivalent to the suicide burn problem - so that might be a helpful keyword to search for. Here’s a stack exchange post with a few solutions to it:


If you have a function for acceleration rather than just a maximum magnitude then that would clarify what you want the graph of velocity to look like.

Assuming a constant acceleration, the max acceleration, one case is that you accelerate up to top speed, continue at top speed for some time, then decelerate to 0 and stop at the precise distance.
d = v*t + a*(v/a)^2 simplifies to t = d/v - v/a, where t is how much time is spent at top speed.

The other case is where the distance is so short that you couldn’t get to top speed and back.
d = a*t^2t = sqrt(d/a), where t indicates how long you accelerate for before decelerating.

1 Like

I don’t really know much about physics but I want to have a realistic acceleration
Would a constant acceleration be realistic? / what kind of function would be realistic?

Edit: btw if possible I would prefer an iterative formula that is based on previous values and delta time but I can probably work with a closed formula too

I would assume that with a top speed, acceleration would slow down before reaching that top speed, like an asymptote. Initial acceleration might also start slow. If you define a function for acceleration then

d = total distance, v = max velocity, a = acceleration function
t = d/v + 2/v*(∫∫a)((∫a)^-1(v)) time spent at max velocity
t = (∫∫a)^-1(d/2) time spent accelerating before decelerating

This requires calculating the inverse of the integral and double integral and inverse of the double integral of the acceleration function.

An iterative formula would turn everything into an approximation.

1 Like

*Is there any parent acceleration function like there is for springs?
Also how would I plug in an initial velocity into the equations if a change is made in the middle?

In my first example the acceleration function is a = max acceleration. If you want something else that better suits whatever your case is then you will need to define some different function for acceleration and calculate the necessary relationships with distance and velocity.

If initial velocity is not 0 then time calculations need to be adjusted; the time needed to accelerate up to the initial velocity needs to be subtracted from the time needed to accelerate to max velocity.

I don’t really understand those equations at all, is there anywhere/anything I should search up to learn about this?

here’s a simulation/demo

local part ="Part")
part.Anchored = true
part.Parent = workspace

local acceleration = 5
local maxvelocity = 32
local distance = 100

local part2 = part:Clone()
part2.CFrame =, -.1, 0)
part2.Parent = workspace


local t, connection = 0
connection = game:GetService("RunService").Heartbeat:Connect(function(dt)
	t = t + dt
	local starttime = maxvelocity/acceleration -- time spent accelerating
	if distance/maxvelocity < starttime then -- will not reach top speed
		local halftime = math.sqrt(distance/acceleration) -- time spent accelerating
		if t >= halftime*2 then
			part.CFrame =, 0, 0)
		elseif t < halftime then
			part.CFrame =*t^2, 0, 0)
			local position = distance/2 + acceleration*halftime*(t-halftime) - acceleration/2*(t-halftime)^2
			part.CFrame =, 0, 0)
		if t >= distance/maxvelocity + maxvelocity/acceleration then
			part.CFrame =, 0, 0)
		elseif t < starttime then
			part.CFrame =*t^2, 0, 0)
			part.CFrame =*(t-starttime/2) - acceleration/2*math.max(0, t-distance/maxvelocity)^2, 0, 0)
1 Like

I’m still confused about the second case

Doesn’t t say how long it will take if you keep accelerating, not how long you accelerate before decelerating?

Also I’m stuck on adding in initial position to this because I think your solution takes advantage of the symmetry between start and end?

Sorry if I’m being really slow xd

In the case where the max speed is not reached, speed is accelerated up to half the distance and decelerated for the other half. The second integral of acceleration with respect to time is the distance traveled; d = (1/2)*a*t^2, but since you also want to decelerate back down to 0, d = (1/2)*a*t^2 + (1/2)*a*t^2 or d = a*t^2, which means t = sqrt(d/a)

If you accelerate for a time of sqrt(d/a) then the distance traveled is (1/2)*a*(sqrt(d/a))^2 = d/2

The second equation of motion says that s = u*t + 1/2*a*t^2 so wouldn’t the entire distance traveled be (1/2)*a*t^2 and a*t^2 would be going double the distance?

a*t^2 is double half the distance because t is explicitly defined as sqrt(d/a)

if these are what the variables represent
d = total distance to travel,
t = how long to accelerate before decelarating (aka 1/2 * (time spent accelerating and decelerating since it is symetrical?)), and
a = the max acceleration

then shouldn’t this be true?
d = 1/2*a*(2*t)^2
d = 2*a*t^2
t = sqrt(d/2/a)

If you accelerate for twice as long then you travel four times the distance. In your case, you accelerate and then decelerate at the same rate for the same amount of time over the same distance.

1 Like

One more quick question, what’s the best way of handling direction? Do you just multiply what you “feel” should be multiplied or is there a systematic way? Should you take abs and sign, or should you use if statements?

If you mean like physically turning something or having some source of propulsion then there are more equations to involve. How something is supposed to happen depends on what it is supposed to be.

I mean just something basic, like for this example but when the distance is negative not positive