Every path begins before the first step.
Every decision, every calculation, every line of code—
starts with a question asked well.
This is the heart of Problem Formulation.
In autonomous systems, robotics, navigation, and control, the hardest part is rarely the solution.
It’s asking the right version of the problem—a version that is clear, solvable, and aligned with reality.
Problem formulation is not paperwork. It is design in its purest form.
It transforms ambiguity into structure.
It draws the boundaries of what matters.
It defines the mission—not just what success looks like, but what must be avoided, respected, or optimized along the way.
A well-formulated problem includes:
– State variables: What do you need to know about the system? Position? Velocity? Heading? Fuel?
– Control inputs: What can you actually change? Steering angle? Thrust? Speed?
– Constraints: What can never be violated? Obstacles, speed limits, actuator bounds, no-fly zones.
– Objective: What are you trying to optimize? Time? Energy? Smoothness? Safety?
– Environment: Is it known or uncertain? Static or dynamic? Flat or curved?
– Initial and terminal conditions: Where does it begin, and where must it end?
In advanced problems, formulation also includes:
– Dynamic models, which describe how the system evolves over time.
– Disturbance models, such as wind, current, or uncertainty.
– Risk tolerance, which defines how safe is “safe enough.”
Formulating the problem is the moment where intuition becomes structure.
It’s the blueprint behind the behavior.
And it’s what separates improvisation from intelligent design.
A poorly formulated problem leads to:
– Overly complex solutions.
– Hidden contradictions.
– Paths that fail in the real world.
A clear formulation, even before any algorithm runs, tells you:
– What’s possible.
– What’s necessary.
– And what matters most.
Because in robotics, as in life, a wrong solution to the right problem can still be useful.
But the right solution to the wrongly formulated problem?
It’s elegant. But irrelevant.
So the discipline begins here—before movement, before control, before optimization.
In that quiet moment where you don’t ask: How do I solve this?
But instead: What is this problem really asking of me?