Proprioceptive Sensing

Encoders are the primary sensors for measuring joint position and velocity. They convert mechanical rotation or linear displacement into digital signals.

• Absolute Encoders: Provide a unique digital code for each shaft position, retaining knowledge of position after power loss. Essential for safety-critical startup.
• Incremental Encoders: Output pulses relative to a starting point, requiring a homing routine. Offer higher resolution and are common in servo motors.
• Key Measurement: Joint Angle (θ) in radians or degrees, and derived Joint Velocity (ω). Resolution is critical for smooth motion and high-gain control.

Force/Torque (F/T) sensors measure the six-dimensional wrench (three forces, three torques) applied at a point, typically mounted at the robot's wrist or in its base.

• Strain Gauge-Based: The most common type; measure minute deformations in a machined element. Provide high bandwidth and accuracy.
• Key Measurements: Contact Forces (Fx, Fy, Fz) and Torques (Tx, Ty, Tz). These are fundamental for:
  • Impedance/Admittance Control: Regulating the robot's dynamic response to contact.
  • Assembly & Insertion: Detecting jams and misalignments.
  • Human-Robot Collaboration: Ensuring safe interaction forces.

Motor current sensing provides an indirect, high-bandwidth measurement of joint torque. The current drawn by a motor is proportional to the torque it generates (τ = k_t * I).

• Hall-Effect Sensors: Non-invasive, measure magnetic field from current-carrying conductor.
• Shunt Resistors: Measure voltage drop across a precision resistor in series with the motor.
• Key Measurement: Motor Current (I). Used for:
  • Torque Control & Limiting: Enabling direct joint torque commands for compliant motion.
  • Fault Detection: Identifying motor stalls, overloads, or winding faults.
  • Gravity Compensation: Calculating the current needed to hold position against gravity.

Inertial Measurement Units (IMUs) are self-contained sensors that measure a system's specific force, angular rate, and orientation.

• Gyroscopes: Measure Angular Velocity (ω) in roll, pitch, and yaw axes.
• Accelerometers: Measure Linear Acceleration (a), which includes both kinematic acceleration and the constant of gravity.
• Key Application: While often exteroceptive for navigation, IMUs provide critical base state estimation for legged robots and mobile manipulators. They help estimate the robot's own body orientation and acceleration, fusing with leg/arm kinematics for stabilization.

While often categorized under touch, tactile sensor arrays provide proprioceptive data about the internal state of contact. They measure the pressure distribution across a sensing surface.

• Technologies: Include capacitive, piezoresistive, and optical (e.g., GelSight) methods.
• Key Measurements: Pressure Map and Contact Geometry. Used for:
  • Slip Detection & Prevention: Sensing incipient object motion within the grasp.
  • Contact State Estimation: Determining how an object is being held (e.g., line contact, surface contact).
  • In-Hand Manipulation: Providing feedback for fine finger adjustments.

Series Elastic Actuators (SEAs) embed proprioceptive sensing intrinsically. A compliant spring is placed between the motor and the output link.

• Core Measurement: Spring Deflection (Δx). Using Hooke's Law (F = k * Δx), this provides a direct, low-noise measurement of Output Torque (τ).
• Key Advantages:
  • High-Fidelity Force Control: The spring filters motor ripple and backlash.
  • Impact Robustness: The spring absorbs and measures shock loads.
  • Energy Storage: Enables efficient dynamic motions like running or jumping.
• This design exemplifies the tight integration of mechanical design and proprioceptive measurement.

Exteroceptive sensing is a robot's ability to perceive its external environment using sensors like cameras, LiDAR, or tactile arrays. It is the complementary modality to proprioception.

• Purpose: Provides information about objects, obstacles, and the world outside the robot's body.
• Key Sensors: RGB-D cameras, stereo vision, time-of-flight sensors, and external tactile sensors.
• Integration: In dexterous manipulation, exteroceptive data (e.g., object pose) is fused with proprioceptive data (e.g., joint torque) to form a complete state estimate for control.

Force-torque sensing measures the multi-axis forces and torques applied at a specific point on a robot, typically at the wrist or in a fingertip. It is a critical subtype of proprioceptive sensing for contact-rich tasks.

• Mechanism: Uses a transducer with strain gauges to measure deformation.
• Applications: Enabling impedance control, detecting contact events, measuring grasp forces, and performing tactile servoing.
• Output: Provides a 6-degree-of-freedom wrench (three forces, three torques).

Joint state estimation is the process of determining a robot manipulator's joint positions, velocities, and sometimes accelerations. It is the most fundamental form of proprioception.

• Primary Sensors: Encoders (optical, magnetic) attached to motors.
• Derived States: Velocity is often computed via numerical differentiation of position; torque/current is measured via motor drivers.
• Role: Essential for computing forward kinematics and the Jacobian matrix, and for implementing low-level joint position or torque control loops.

A Series Elastic Actuator (SEA) is a robotic actuator that incorporates a compliant element (like a spring) in series with the motor. This design enables high-fidelity proprioceptive force sensing and control.

• Principle: The spring deflects under load; measuring this deflection (with an encoder) provides a direct, low-noise measurement of output force.
• Benefits: Accurate force control, inherent shock absorption, and safety in human-robot interaction.
• Trade-off: Introduces mechanical bandwidth limitations compared to rigid actuators.

Gravity compensation is a model-based control technique that calculates and commands the joint torques needed to counteract the weight of a robot's own links. It relies on accurate proprioceptive joint state data and a dynamic model.

• Purpose: Allows the robot to move its arm as if in zero gravity, making it easier to control for fine manipulation or to measure external forces.
• Implementation: Uses the robot's dynamic model and real-time joint angle feedback to compute the required torque vector.
• Prerequisite: Accurate calibration of link masses, centers of mass, and inertial parameters.

Tactile servoing is a closed-loop control method that uses real-time tactile sensor feedback to guide robotic manipulation. It depends on high-bandwidth proprioceptive-like signals from the contact interface.

• Feedback Signal: Uses data from tactile arrays (e.g., GelSight) or force-torque sensors.
• Objective: Maintain a desired contact state, such as a specific force profile, or follow a contour on an object's surface.
• Relation to Proprioception: While tactile sensing is often exteroceptive, its use in a tight servo loop for direct motor control makes it functionally analogous to high-resolution proprioception at the end-effector.