Control Theory (PID / LQR / MPC)
Making the robot actually do what you want.
Why it matters in robotics
Control is the layer that turns a planned trajectory or desired state into stable, executable motor commands, so robotics interviews lean on it heavily for anything that moves: arms, drones, legged robots, and self-driving stacks. Expect to be asked to tune or reason about a PID loop, explain when LQR's optimality and gain matrix beat hand-tuning, and articulate why MPC is worth its compute cost when constraints (joint limits, obstacles, actuator saturation) matter. Strong candidates can connect the math (poles, stability, cost functions) to practical failure modes like integrator windup, steady-state error, and model mismatch.
Application focus
The same topic, tailored to the robot you're building. Your choice is remembered across the roadmap and every topic.
At a glance
The closed feedback loop common to PID, LQR, and MPC: the controller drives the plant so its measured state tracks the reference, using the error between them.
What to study
- ✓PID fundamentals: P/I/D terms, tuning intuition, steady-state error, and integrator anti-windup
- ✓State-space modeling and stability: poles/eigenvalues, controllability/observability, pole placement
- ✓LQR as optimal full-state feedback: the quadratic cost, Q/R weighting tradeoffs, and the Riccati equation
- ✓MPC: receding-horizon optimization, handling state/input constraints, and the cost of online computation
Study by time budget
Pick the path that fits the time you have before your interview.