The controller is the "brain" of the robot, a dedicated computer system that orchestrates the entire operation. It bridges the gap between the high-level task plan and the low-level physical motion of the manipulator. Its primary function is to execute a control law that ensures the robot accurately follows a desired trajectory.

The Control Loop

The fundamental operation of a robot controller is the feedback control loop, a cycle that typically repeats at a high frequency (often 1 kHz or higher).

Sense

The controller reads the current, actual state of the robot from its proprioceptive sensors (e.g., the angles of all joints from their encoders).

Compare

It compares this actual state to the desired state for that specific instant in time, as dictated by the pre-computed trajectory. The difference between the desired and actual state is the error.

Act

Based on the calculated error, the controller's control law (a specific algorithm, such as a PID controller) computes the necessary torque or voltage command for each joint's actuator. This command is sent to the motor amplifiers, which drive the motors to move the robot in a way that minimizes the error.

This continuous process of sensing, comparing, and correcting is what allows a robot to track a trajectory smoothly and accurately, even in the presence of disturbances like varying loads or friction.

Controller Architecture

A typical industrial robot controller is a complex piece of hardware and software with a hierarchical structure.

Hardware

The controller is housed in a cabinet that contains:

Main Computer

A powerful processor (or multiple processors) that runs the high-level control software, including trajectory planning, kinematics, and dynamics calculations.

Memory

Stores the robot's operating system, application programs, and configuration data.

I/O (Input/Output) Boards

Provide the interface to communicate with the robot's sensors, actuators, and external equipment (like safety systems or other machines).

Motor Amplifiers (Drives)

These are power electronics that take the low-voltage control signal from the main computer and convert it into the high-current power needed to drive the robot's motors. Each motor on the robot has a corresponding amplifier.

Teach Pendant

A handheld device with a screen and joystick that allows a human operator to program the robot, manually move it (a process called "jogging"), and monitor its status.

Software

The software architecture is also hierarchical, with different tasks operating at different rates.

High-Level Control (Task Level)

This level is responsible for user interaction and interpreting the robot program. It plans the overall task and generates the desired path for the end-effector in Cartesian space (e.g., "move in a straight line from point A to point B").

Mid-Level Control (Trajectory Level)

This layer takes the desired Cartesian path and transforms it into a time-sequenced trajectory of points. It then uses the robot's inverse kinematics model to calculate the corresponding sequence of desired joint angles, velocities, and accelerations for each point in time. For more advanced control, it may also use the robot's dynamic model to calculate the feedforward torques needed to achieve that motion.

Low-Level Control (Servo Level)

This is the fastest and most fundamental layer, running at the highest frequency (e.g., 1-10 kHz). It runs the actual joint control loops, taking the desired joint setpoints for each instant from the trajectory level and commanding the motor amplifiers to achieve them using a feedback control law like PID.