Hand Guiding

The robot's control system actively counteracts the effects of gravity and joint friction, allowing the operator to move the arm with minimal effort. This is achieved through dynamic models and torque control at each joint.

• Key Mechanism: Uses an internal dynamic model to calculate the required joint torques to hold the arm's current position against gravity.
• User Experience: The arm feels weightless and responsive, enabling precise positioning without operator fatigue.
• Technical Requirement: Depends on accurate robot mass/inertia parameters and low-friction transmissions.

Hand guiding serves as a primary method for kinesthetic teaching or Programming by Demonstration (PbD). The operator physically moves the robot through a desired task sequence, with the system recording key waypoints, trajectories, and orientations.

• Process: The operator typically holds a dedicated teach handle or the robot's end-effector, presses an enabling device (dead-man switch), and guides the motion. Waypoints are recorded via a button press.

• Output: Creates a robot program consisting of precise position/orientation targets, often with associated motion parameters (speed, blending).

• Advantage: Eliminates the need for traditional joystick or code-based programming, making automation accessible to non-experts.

The robot operates in an admittance control mode, where an applied external force (from the human) is interpreted as a desired velocity or position change. The control law makes the robot behave like a spring-mass-damper system.

• Control Loop: Measures forces/torques (via joint current sensors or a six-axis force/torque sensor at the wrist), computes a desired motion, and commands the motors to achieve it.

• Parameter Tuning: Stiffness and damping parameters can be adjusted to make the robot feel 'soft' and easy to move or 'stiff' for more precise guidance.

• Contrast with Impedance Control: Admittance control (force in, motion out) is typically used for hand guiding, whereas impedance control (motion in, force out) is used for interaction with rigid environments.

Hand guiding is a collaborative operation mode defined in safety standards like ISO/TS 15066. Its implementation incorporates multiple safety layers:

• Enabling Device: A required three-position dead-man switch that must be partially depressed to allow motion. Releasing or fully depressing it triggers a protective stop.

• Force and Speed Monitoring: The system continuously monitors commanded and actual forces/velocities. If thresholds are exceeded (e.g., due to a collision), a safety-rated stop is triggered.

• Safe Limited Speed: The maximum guided speed is capped to a safe level (often ≤ 250 mm/s as per standards) to limit kinetic energy.

• Workspace Monitoring: Can be integrated with external safety sensors (laser scanners, light curtains) to define collaborative workspaces.

Advanced hand guiding systems can overlay virtual fixtures—software-defined guidance geometries or constraints—onto the physical motion. These fixtures assist the operator by providing haptic feedback or limiting movement.

• Types of Fixtures:

  • Guidance Virtual Fixtures: Attractive forces that channel the tool along a desired path (e.g., a straight line for a weld seam).
  • Forbidden Region Virtual Fixtures: Repulsive forces or hard stops that prevent movement into restricted volumes (e.g., near a fragile part).

• Implementation: Requires accurate robot and environment models. Fixtures are rendered by modulating the control law based on the tool's position relative to the virtual geometry.

• Benefit: Dramatically improves precision and reduces cognitive load during complex manual tasks like assembly or polishing.

A robust hand guiding system allows fluid transitions between different operational modes, enabling a flexible workflow.

• Typical Transition Path:

  1. Automatic Mode: Robot executes a pre-programmed task.
  2. Trigger: Operator initiates hand guiding (e.g., grabs handle, presses button).
  3. Transition: Control system smoothly switches from position/velocity control to admittance control without jerks or overshoot.
  4. Hand Guiding Mode: Operator manually repositions the robot.
  5. Return to Auto: Operator confirms new position, system transitions back to automatic execution from the updated pose.

• Technical Challenge: Requires stable, bumpless transfer between fundamentally different control paradigms to ensure safety and smooth motion.

Admittance Control is the underlying control scheme that enables compliant Hand Guiding. It allows the robot to move in response to an external force. The core mechanism involves:

1. Force/Torque Sensing: Measuring the applied human force via sensors in the robot's joints or wrist.
2. Desired Dynamics Calculation: Using a virtual model (mass-spring-damper system) to convert the sensed force into a desired motion or velocity.
3. Motion Execution: The robot's inner position or velocity controller drives the motors to achieve this compliant motion. This creates the feeling of moving a lightweight, damped object. The admittance parameters (virtual mass, damping) can be tuned to make the robot feel heavy and stable or light and responsive.

Shared Autonomy is a broader control paradigm where control authority is dynamically blended between a human operator and the robot's autonomous intelligence. Hand Guiding represents one point on this spectrum—full human manual control. In a Shared Autonomy system, the robot might:

• Interpret the human's intent from partial guidance and complete the task.
• Apply virtual fixtures to keep the guided motion within safe or precise bounds.
• Filter human inputs to reject tremors or correct for poor alignment.
• Take over for certain sub-tasks (e.g., precise insertion) before returning control. This framework allows for smooth transitions between direct physical guidance (Hand Guiding) and assisted or fully autonomous operation.

Physical Human-Robot Interaction (pHRI) is the academic and engineering field encompassing all scenarios where humans and robots make deliberate physical contact. Hand Guiding is a canonical example of intentional, collaborative pHRI. The field's research focuses on:

• Safe control architectures: Ensuring stability and safety during force exchange.
• Intuitive physical interfaces: Designing handles and grips for effective guidance.
• Human biomechanics: Understanding injury thresholds and comfortable force levels.
• Dynamic role adaptation: How a robot can sense and respond to the human's physical actions beyond simple guidance. pHRI differentiates itself from Social HRI by its focus on force and physical collaboration rather than purely social communication.