Sensors & Actuators

What it is

Sensors let a robot perceive its environment (cameras, LiDAR, IMUs, encoders); actuators let it act on the physical world (motors, servos) — together, they're the hardware interface all robotics software ultimately reads from and writes to.

Key points

• Sensor noise is the default, not the exception: every real sensor reading has noise and uncertainty — robotics software is built around filtering and estimation (e.g. Kalman filters) specifically because raw sensor data can't be trusted at face value.
• Proprioceptive vs. exteroceptive sensors: proprioceptive sensors (encoders, IMUs) measure the robot's own internal state (joint angles, orientation); exteroceptive sensors (cameras, LiDAR) measure the external environment — both are needed for a robot to know both where it is and what's around it.
• Actuator control loops: actuators are rarely commanded open-loop — a feedback control loop (commonly PID) continuously corrects for the difference between commanded and actual position/velocity.
• Sensor and actuator choice drives most downstream software architecture decisions — the perception and control stack has to be built around what the hardware can actually provide and do.