Ground Reaction Force (GRF)

The Normal Force is the component of the GRF vector that acts perpendicular to the contact surface. It is a direct consequence of the robot's weight and any vertical acceleration.

• Primary Role: Supports the robot against gravity and provides the necessary friction for propulsion.
• Measurement: Typically the largest magnitude component during level-ground walking.
• Control Implication: Insufficient normal force leads to foot slippage, while excessive force can indicate inefficient or unstable gait dynamics. In Whole-Body Control, it is a key constraint to prevent loss of contact.

The Shear Forces are the tangential components of the GRF vector parallel to the ground plane. They are decomposed into two orthogonal directions:

• Anterior-Posterior (Fore-Aft): Acts along the direction of travel. A positive (forward) shear propels the robot, while a negative (backward) shear brakes its motion.
• Medial-Lateral (Side-to-Side): Acts perpendicular to the direction of travel. This component is critical for lateral stability, especially during turning maneuvers or on sloped terrain.

These forces are generated by friction and are essential for generating acceleration and resisting external pushes.

The Center of Pressure is not a force component but the point of application of the resultant GRF vector on the contact surface.

• Definition: The single point where the total moment of the distributed pressure field is zero.
• Stability Significance: The location of the CoP relative to the Support Polygon is the primary indicator of static and dynamic stability. For a robot to remain balanced without tipping, the CoP must remain within the convex hull of its contact points.
• Measurement: Calculated from force/torque sensors in the foot or ankle. Its trajectory is a key signal for Push Recovery and balance controllers.

The Resultant GRF is the vector sum of all three orthogonal components. Its overall magnitude and spatial orientation encapsulate the total mechanical interaction with the ground.

• Magnitude: Calculated as √(Fx² + Fy² + Fz²). Its profile over a gait cycle (a "force curve") is a signature of the locomotor strategy (e.g., running has higher, sharper peaks than walking).
• Direction: The 3D angle of the resultant vector. In stable walking, it generally points slightly ahead of and through the robot's Center of Mass, creating a moment that propels the body forward while supporting it.

The GRF is a dynamic signal that evolves over the duration of foot contact. Analyzing its temporal profile is as important as its spatial components.

• Gait Cycle Analysis: The characteristic shape of the force-time graph differs for walking (double-hump) vs. running (single sharp peak).
• Impulse: The integral of the GRF vector over time. The Anterior-Posterior Impulse equals the change in forward momentum. The Vertical Impulse equals the change in vertical momentum and must counteract gravity over the stride.
• Application: Used in Inverse Dynamics to calculate net joint torques and in evaluating the Cost of Transport.

The net GRF from all contact points directly governs the motion of the robot's Center of Mass according to Centroidal Dynamics.

• Newton-Euler Equations: The sum of all external GRFs equals the total mass times CoM acceleration (ΣF = m*a_com). The sum of GRF moments about the CoM equals the rate of change of Centroidal Angular Momentum.
• Planning & Control: This relationship is foundational for high-level planners. The Linear Inverted Pendulum Model simplifies this by assuming the GRF vector always points from the CoP toward the CoM. Model Predictive Control uses these dynamics to optimize future footstep placements and CoM trajectories.

The Center of Pressure (CoP) is the specific point on a contact surface where the total Ground Reaction Force (GRF) vector is considered to act. It is the centroid of the distributed pressure field under a foot. For stability, the CoP must remain within the support polygon. If the CoP reaches the polygon's edge, the robot begins to tip, and the GRF can no longer generate a counter-moment to prevent a fall.

• Key Relationship: The CoP location is directly determined by the distribution of the normal component of the GRF.
• Measurement: Often estimated using force-sensitive resistors (FSRs) or multi-axis force/torque sensors in the foot.
• Control Objective: Many balance controllers, like those based on the Zero-Moment Point, explicitly regulate foot pressure to manipulate the CoP.

The Zero-Moment Point (ZMP) is a theoretical point on the ground where the net moment of all inertial and gravitational forces acting on the robot has zero horizontal component. It is a dynamic stability criterion used extensively for bipedal walking.

• Core Principle: For a robot to walk without rotating about its edges (i.e., without falling), the ZMP must lie within the support polygon.
• Connection to GRF: In most practical analyses on rigid, level ground, the ZMP is equivalent to the Center of Pressure (CoP). The horizontal moments are zero precisely where the GRF is applied.
• Planning Use: ZMP-based gait generation pre-plans footstep locations and Center of Mass trajectories to ensure the ZMP stays within a stable region.

Centroidal dynamics governs the motion of a robot's Center of Mass (CoM) and the evolution of its centroidal angular momentum (the angular momentum about its CoM). It provides a high-level description directly linked to the net Ground Reaction Force.

• Fundamental Equation: The rate of change of the robot's total linear momentum equals the sum of external forces (primarily gravity + GRF). The rate of change of centroidal angular momentum equals the net external moment about the CoM.
• Planning & Control: Whole-body controllers use centroidal dynamics to compute the required net GRF and moment to achieve a desired CoM acceleration. The solver then distributes this net wrench across the individual foot GRFs.
• Simplification: Models like the Linear Inverted Pendulum Model (LIPM) are derived from specific assumptions about centroidal dynamics (e.g., constant height, zero angular momentum).

Whole-Body Control (WBC) is a hierarchical optimization framework that coordinates all of a robot's joints to execute multiple tasks simultaneously (e.g., foot placement, torso orientation, arm manipulation) while respecting physical constraints like GRF limits and friction cones.

• GRF as Control Output: A primary output of a WBC solver is the optimal set of Ground Reaction Forces at each contact point needed to achieve the task-level objectives and maintain balance.
• Constraint Integration: WBC explicitly enforces that GRFs must:
  • Lie within friction cones to prevent slipping.
  • Provide only push forces (no adhesion).
  • Sum to the required net wrench for centroidal dynamics.
• Real-Time Execution: Typically formulated as a Quadratic Program (QP) solved at high frequency (e.g., 1 kHz).

Model Predictive Control (MPC) is an advanced control method that uses an internal dynamic model to predict the robot's future state over a time horizon. It solves an optimization problem at each control cycle to find the optimal sequence of future control inputs, with only the first step applied.

• GRF as Decision Variable: In legged locomotion MPC, the future Ground Reaction Forces are often the primary decision variables. The optimizer selects GRFs that minimize error (e.g., in CoM trajectory) while satisfying dynamics and constraints.
• Anticipatory Action: By looking ahead, MPC can pre-emptively adjust GRF profiles to handle upcoming disturbances or terrain changes, enabling more dynamic and robust locomotion than reactive controllers.
• Common Model: The Linear Inverted Pendulum Model (LIPM) is frequently used within MPC for its computational efficiency, with the GRF directly linked to CoM acceleration.

The Spring-Loaded Inverted Pendulum (SLIP) model is a reduced-order model that captures the essential dynamics of running, hopping, and dynamic walking. It represents the leg as a massless, compliant spring attached to a point-mass body.

• GRF Generation Mechanism: In the SLIP model, the Ground Reaction Force is not directly controlled but emerges from the passive compression and extension of the spring leg during ground contact. The force profile is sinusoidal.
• Energy Efficiency: It models the passive exchange between kinetic and potential energy (via the spring), explaining the natural dynamics and low Cost of Transport (CoT) observed in animals and some robots.
• Template for Control: Many control strategies for advanced legged robots (e.g., Boston Dynamics' Atlas) use SLIP-inspired concepts to generate natural, dynamic gaits where GRF profiles are shaped by virtual leg compliance.