The previous post in this neighborhood — A Field Guide to Path Planning — covered how a robot decides where to go. This one picks up the next layer: given a path the planner produced, how does the robot actually follow it?
That’s a different problem than people new to robotics often expect. A path is just a curve in space. Following it is a control problem. And it’s a different control problem depending on what kind of robot you have — a wheeled vehicle has nonholonomic constraints, an arm has redundant degrees of freedom and joint limits, a quadcopter has six-DoF dynamics and underactuation. Each one gets its own family of techniques. This primer walks through all three side by side.
What it covers
Seven sections, eight live demos, about thirty minutes to read.
§1 — Three layers of robot motion. The pyramid: planning (where), tracking (how), control (the actuator commands). Why each layer exists and where most engineering time actually goes.
§2 — Path tracking for wheeled vehicles. Four demos: pure pursuit (the classic, with its single tunable lookahead), the Stanley controller (the geometric improvement that won the DARPA Grand Challenge), kinematic MPC (the modern workhorse), and a side-by-side that shows why each one wins on different geometries.
§3 — Arm navigation. Two demos. The inverse kinematics question — given a desired end-effector pose, what joint angles get me there? — including the redundancy and singularity problems that make it harder than it looks. Then RRT-Connect: the sampling-based planner that turned arm motion planning from “academic” to “shipped on every modern manipulator.”
§4 — Aerial navigation. Two demos. The cascaded controller architecture every drone runs (position → attitude → motor speeds, fast inner loop slow outer loop). Then geometric tracking on SE(3), the approach that handles aggressive maneuvers cleanly without singularities.
§5–7 — Side by side, practical engineering notes, references. Comparison across all three robot types. The engineering questions that actually decide algorithm choice in production. A short reading list of the papers and books worth knowing.
Every algorithm has a worked example, and the eight live demos let you drag goals, change parameters, and watch the controllers respond in real time.
Read it
The primer is a natural follow-on to the path planning field guide — same visual language, same level of technical depth, picking up exactly where the previous one left off.
← Back to Autonomy