Impedance Control

Unlike position control which commands a specific trajectory, or force control which commands a specific contact force, impedance control regulates the dynamic relationship between the two. It defines a desired mechanical impedance—modeled as a mass-spring-damper system—at the robot's end-effector. When the robot contacts an object, the resulting interaction force causes a deviation from the commanded position according to this defined relationship. This allows the robot to behave as if it were a programmable physical object with specific compliance properties.

The desired behavior is mathematically defined by three key parameters:

• Stiffness (K): The spring constant. High stiffness resists positional deviation, behaving like a rigid tool. Low stiffness allows the end-effector to "give" under force, useful for delicate tasks.
• Damping (B): The damping coefficient. This dissipates energy, preventing oscillations upon contact and ensuring stable, smooth interaction.
• Inertia (M): The apparent mass. This defines how the end-effector accelerates in response to net forces. By tuning these parameters, engineers can make a robot feel heavy and sluggish, light and responsive, or stiff and precise, depending on the task.

A primary advantage of impedance control is its robustness in unstructured environments where contact geometry and timing are uncertain. Because it does not demand perfect tracking of a pre-planned position trajectory (which would cause large, potentially damaging forces upon unexpected contact), it is more forgiving. The controller naturally accommodates surface variations and minor misalignments, making it ideal for tasks like compliant assembly (e.g., inserting a peg into a hole with tight tolerances), polishing, or physical human-robot interaction where safety is paramount.

Impedance control is often contrasted with its dual, admittance control. The key distinction lies in the causality of the control loop:

• Impedance Control: Position in, force out. The controller accepts a position command and, using torque control at the joints, makes the robot behave like the desired impedance. Force is a measured output.
• Admittance Control: Force in, position out. The controller uses a force/torque sensor to measure external contact forces. It then computes a position adjustment according to a desired admittance (the inverse of impedance) and sends this to a high-gain position controller. Impedance control typically offers better high-frequency force response but requires accurate joint torque control. Admittance control is often easier to implement on standard industrial robots with position interfaces.

Successful implementation relies on an accurate dynamic model of the manipulator. The controller must compensate for the robot's own inherent dynamics—such as link inertia, Coriolis forces, and gravity—to ensure the perceived impedance at the end-effector matches the desired, user-defined parameters. Any inaccuracy in this model results in a distorted impedance. For example, unmodeled gravity compensation would make the arm feel heavier in certain postures. Advanced implementations use adaptive control or online parameter estimation to maintain consistent impedance across the robot's workspace and with varying payloads.

By setting low stiffness and appropriate damping, impedance control is a foundational technology for collaborative robots (cobots). It allows the robot to safely yield to human contact, either intentionally (for guided teaching) or accidentally. This inherent compliance reduces the risk of injury and damage. In applications like hand-guiding, the human operator can physically move the robot arm, and the impedance controller makes it feel light and responsive. This characteristic is critical for deploying robots in shared workspaces without traditional safety cages.

Admittance control is the dual strategy to impedance control. Instead of regulating the force resulting from a position error, it regulates the position resulting from a measured force. The controller takes a measured force/torque from a sensor and computes a commanded motion, effectively specifying how compliant the robot should be. This architecture is often implemented in stiff, position-controlled industrial robots by adding an outer control loop that modifies the position setpoint based on force feedback.

• Key Distinction: Impedance control commands torque; admittance control commands motion.
• Typical Hardware: Requires a six-axis force/torque sensor at the wrist.
• Use Case: Ideal for collaborative assembly tasks where the robot must 'give way' to human contact or part misalignment.

Force/torque (F/T) sensing is the direct measurement of the multi-dimensional forces and torques applied at a robot's wrist or end-effector. This data is the critical sensory input for both impedance and admittance control loops. A six-axis F/T sensor measures three linear forces (Fx, Fy, Fz) and three rotational torques (Tx, Ty, Tz).

• Implementation: Sensors use strain gauges or optical principles to detect minute deflections.
• Applications: Enables compliant polishing, delicate part insertion, and human-robot contact detection.
• Challenge: Signals are noisy and must be filtered; sensor data must be transformed from the sensor frame to the robot's control frame.

Compliant assembly is a class of robotic tasks, such as peg-in-hole insertion or screw driving, where precise part mating is achieved by allowing the robot to accommodate positional uncertainties through controlled compliance. Impedance control is a primary method to implement this, setting a low stiffness in the lateral directions so contact forces gently guide the peg into the hole.

• Methods: Can be achieved via active compliance (impedance/admittance control) or passive compliance (using mechanical devices like Remote Center of Compliance units).
• Key Parameter: Center of compliance must be aligned with the contact point for stable guidance.
• Industry Standard: Essential for automotive and electronics manufacturing where part tolerances are tight.

Model Predictive Control (MPC) is an advanced optimization-based control strategy that can incorporate impedance objectives. An MPC controller uses an internal dynamic model to predict the robot's future states over a finite horizon and solves for a sequence of optimal control inputs that minimize a cost function, which can include terms for tracking a desired impedance or limiting contact forces.

• Advantage over Impedance Control: Can explicitly handle actuator limits, obstacle constraints, and complex multi-objective tasks.
• Computational Demand: Requires solving an optimization problem in real-time, demanding significant processing power.
• Hybrid Approach: Modern systems often use MPC to plan force trajectories that a local impedance controller then tracks.

Whole-Body Control (WBC) is a hierarchical control framework for complex robots (e.g., humanoids, mobile manipulators) that coordinates all joints to execute multiple tasks simultaneously. Impedance control can be formulated as one of several task-space objectives within a WBC framework. For example, a humanoid could use WBC to maintain balance (primary task) while its arm executes an impedance-controlled pushing task (secondary task).

• Core Mechanism: Uses null-space projection to achieve lower-priority tasks without affecting higher-priority ones.
• Integration: Desired impedance at an end-effector becomes a quadratic cost term in the WBC optimization.
• Application: Critical for robots that must interact with the environment while managing their own stability.

Teleoperation is the direct, real-time remote control of a robotic manipulator by a human operator. Bilateral teleoperation systems use impedance or admittance control principles to create force feedback, allowing the operator to feel the contact forces experienced by the remote robot. This creates a transparent haptic interface.

• Architecture: The remote robot (slave) often runs an impedance controller to render its environment. The local interface (master) measures the human's force and runs an admittance controller to provide kinesthetic feedback.
• Challenge: Maintaining stability despite time delays in communication networks.
• Use Case: Essential for surgery, underwater exploration, and handling hazardous materials.