Force Closure

A grasp achieves force closure when the set of possible contact wrenches applied at the contact points positively spans the entire six-dimensional wrench space. This means that for any external wrench (combined force and torque) acting on the object, the robot can generate a set of feasible contact forces to exactly oppose it, holding the object immobile.

• Key Condition: The convex hull of the primitive contact wrenches must contain the origin of the wrench space in its interior.
• Result: The object is immobilized and cannot twist or translate, regardless of the direction of the disturbance.

Force closure and form closure are both grasp stability concepts, but they differ fundamentally in their reliance on friction.

• Form Closure: Achieves immobilization through geometric constraints alone, with frictionless contacts. The object cannot move even if contact forces are zero. This typically requires at least 7 frictionless contact points in 3D.
• Force Closure: Relies on friction to generate tangential forces at the contacts. This allows secure grasping with far fewer contact points (e.g., a two-fingered pinch grasp can achieve force closure but not form closure).
• Practical Implication: Almost all dexterous robotic grasps in the real world rely on force closure, as form closure's geometric requirements are too restrictive.

The friction cone is the central geometric model enabling force closure analysis. At each contact point, the set of allowable contact forces is limited by Coulomb friction.

• Definition: A friction cone is the set of all possible contact force vectors that satisfy the friction constraint |F_tangential| ≤ μ * |F_normal|, where μ is the coefficient of friction.
• Visualization: A cone whose axis is aligned with the surface normal at the contact point.
• Analysis: For force closure, the wrench cones (the friction cones mapped into wrench space via the grasp matrix) must positively span the wrench space. The larger the friction coefficient μ, the wider the cones, making force closure easier to achieve.

Not all force closure grasps are equally good. Quality metrics quantify robustness.

• Ferrari-Canny Metric (ε): Measures the radius of the largest sphere, centered at the origin, that can be inscribed within the convex hull of the primitive wrenches in the grasp wrench space. A larger ε indicates a grasp can resist larger external disturbances.
• Margin of Stability: The minimum factor by which contact forces can be scaled down before the grasp loses force closure.
• Task-Directed Metrics: Measure the grasp's ability to resist wrenches expected for a specific task (e.g., lifting against gravity, resisting a turning torque).

For a grasp to potentially achieve force closure, certain basic geometric and combinatorial conditions must be met.

• Number of Contacts: In 3D, at least 4 hard-finger contacts with friction are theoretically sufficient, but 7 are required for frictionless form closure.
• Contact Geometry: Contacts must be arranged such that their corresponding wrench vectors are not all contained within a single half-space of the wrench space.
• Force Feasibility: The required contact forces to balance external wrenches must lie within the friction cones and not exceed the actuators' force limits.

Force closure theory directly informs the algorithms used to plan robotic grasps.

• Search Objective: Grasp planners like those in DexNet often use a force closure quality metric (e.g., the Ferrari-Canny metric) as the primary objective to optimize when evaluating candidate grasp poses from point clouds.
• Robustness to Uncertainty: A high-quality force closure grasp provides a margin of error for uncertainties in object pose estimation, contact point location, and friction coefficients.
• Integration with Control: The existence of force closure validates that a grasp wrench space exists, which can be used by underlying impedance or force control strategies to maintain stability during manipulation.

Force closure is the gold-standard analytical metric for evaluating the robustness of a candidate grasp. It is quantified by analyzing the Grasp Wrench Space (GWS)—the set of all wrenches a grasp can apply. A grasp achieves force closure if the GWS contains the origin of the wrench space in its interior, meaning it can resist disturbances in any direction. This is often computed via the Ferrari-Canny metric, which measures the radius of the largest wrench sphere centered at the origin that fits inside the GWS. A larger radius indicates a more robust, disturbance-resistant grasp.

For simple parallel-jaw grippers, achieving force closure requires careful consideration of contact geometry and friction. Key principles include:

• Form Closure vs. Force Closure: A parallel-jaw grasp on a rectangular prism may only achieve form closure (geometric restraint) if the jaws perfectly encase it. Force closure adds the ability to apply squeezing forces to resist pulls.
• Friction Cones: The grasp must be arranged so that the friction cones at the contact points, which represent all possible contact forces due to friction, collectively span the wrench space.
• Example: Gripping a cylinder. A grasp with the jaws tangent to the sides (line contact) cannot generate torque about the cylinder's axis, failing force closure. A grasp that encloses the cylinder, creating opposing contact normals, can achieve it.

Multi-fingered, dexterous hands (e.g., Shadow Hand, Allegro Hand) use force closure as a primary criterion for in-hand manipulation and stable prehensile tasks. Engineers use it to:

• Plan Grasp Configurations: For a given object model, algorithms search for finger contact points that collectively satisfy force closure conditions.
• Evaluate Underactuation: In underactuated hands, where fingers conform to object shape, force closure analysis ensures the passive adaptation still produces a stable equilibrium.
• Dynamic Tasks: For actions like using a tool, the required wrench set changes. Force closure analysis ensures the grasp can provide the specific forces and torques needed for the task, not just static stability.

Beyond grasping, force closure principles are essential in industrial fixturing—designing clamps and vices to hold workpieces during machining. The goal is to immobilize the part completely against cutting forces. This involves:

• Strategic Contact Placement: Locators and clamps are positioned to ensure the composite wrench space of all contacts can counteract any machining force.
• Frictionless Analysis: Often, a conservative frictionless force closure analysis is performed, assuming zero friction at contacts. If closure is achieved under this assumption, the fixture is extremely robust.
• 3-2-1 Locating Principle: This classic manufacturing rule for constraining 6 degrees of freedom is a practical application of form/force closure theory, ensuring deterministic part placement.

Modern data-driven grasp synthesis systems, like DexNet, use force closure as a core training signal. In simulation:

• Millions of grasp candidates are generated on object meshes from datasets like the YCB Object Set.
• Each candidate is evaluated by computing its force closure metric (e.g., Ferrari-Canny score) under simulated physical properties (friction, object mass).
• A deep neural network is then trained to predict this quality score from visual point clouds, learning to propose grasps that are analytically robust. This bridges the sim-to-real gap by optimizing for a fundamental physical property.

Force closure is a necessary but not always sufficient condition for a successful real-world grasp. Practitioners must consider:

• Soft Finger Contacts: The standard model assumes hard, point contacts. Soft, deformable fingertips (e.g., using GelSight sensors) create contact patches that can exert torsional moments, modifying the wrench space.
• Dynamic Stability: Force closure analyzes static equilibrium. A grasp may be statically stable but fail during dynamic manipulation due to inertial forces.
• Task Compatibility: A force-closure grasp might place fingers in positions that obstruct the intended task (e.g., blocking a screwdriver's path). This leads to task-oriented grasping, where force closure is one constraint among many.
• Tactile Feedback: Maintaining force closure in the face of disturbances relies on tactile servoing and slip detection to actively adjust grip forces.

The Grasp Wrench Space (GWS) is the set of all possible wrenches (combined forces and torques) that can be applied to an object by a robotic grasp through its contact points. It is the fundamental geometric construct used to analyze grasp stability.

• Mathematical Basis: Each contact point can apply a force within a friction cone. The GWS is the convex hull of all wrenches resulting from forces within these cones.
• Relation to Force Closure: A grasp achieves force closure if the GWS contains a neighborhood around the origin, meaning it can resist any external wrench.
• Practical Use: Engineers compute approximations of the GWS to evaluate and score candidate grasps for robustness before execution.

Form closure is a stricter, geometric condition for grasp immobilization where an object is held securely by rigid contacts alone, without relying on friction. It is a purely kinematic constraint.

• Key Difference from Force Closure: Form closure does not require friction. The object's motion is prevented by the geometry of the contacts themselves. All force closure grasps are form closure grasps if friction is ignored.
• Mathematical Test: Form closure exists if the only solution to the equilibrium equations, with contact forces directed inward, is the zero vector.
• Application: Crucial for designing fixtures in manufacturing and for grasps in high-vibration or low-friction environments.

A friction cone is the geometric representation of all possible contact force vectors that satisfy the Coulomb friction model at a point of contact between two bodies.

• Definition: For a given contact normal and coefficient of friction (μ), the cone contains all force vectors whose tangential component does not exceed μ times the normal component. This ensures the force does not cause slip.
• Role in Grasp Analysis: The wrench space of a grasp is built from the set of forces inside the friction cones at each contact. A wider cone (higher μ) or more contacts expands the possible wrench space, making force closure easier to achieve.
• Linear Approximation: In computational analysis, the nonlinear cone is often approximated as a polyhedral cone for linear programming solutions.

Impedance control is a robot control strategy that regulates the dynamic relationship between force and motion at the end-effector, making the robot behave as a programmable mass-spring-damper system. It is essential for maintaining stable contact during force-closure grasps.

• Mechanism: Instead of directly commanding force, it commands an impedance (stiffness, damping, inertia). The controller modulates motion in response to measured contact forces: F = K*x + B*ẋ + M*ẍ.
• Application to Grasping: Allows a hand to comply with object geometry during grasp acquisition, ensuring even force distribution and avoiding excessive contact forces that could break objects or destabilize the grasp.
• Contrast with Admittance Control: Impedance control accepts position commands and outputs force. Admittance control (its dual) accepts force measurements and outputs motion commands.

Tactile servoing is a closed-loop control method that uses real-time tactile sensor feedback to guide robotic manipulation, such as maintaining grip force or adjusting finger placement to prevent slip.

• Purpose: To achieve and maintain force closure under real-world uncertainties like object deformation, vibrations, or external disturbances.
• Process: Tactile arrays (e.g., GelSight, BioTac) provide high-resolution pressure or shear maps. The controller uses this feedback to adjust joint torques or positions, servoing on tactile features like contact centroid or pressure distribution.
• Key for Dexterity: Enables in-hand manipulation and regrasping without vision by feeling the object's motion within the grasp, directly addressing the stability requirements of force closure.

Contact-implicit trajectory optimization is a planning method that optimizes robot motions without pre-specifying the sequence or timing of contacts, allowing the solver to discover when and where contacts should occur to achieve a goal.

• Relation to Force Closure: This method can automatically synthesize complex manipulation trajectories that involve making and breaking force-closure grasps, such as flipping or rolling an object.
• Mathematical Approach: Formulates the problem with complementarity constraints that model the discontinuous nature of contact (e.g., no inter-penetration, forces only when in contact). The solver finds states, controls, and contact forces simultaneously.
• Advantage: Eliminates the need for manually scripting contact modes, enabling the discovery of dynamic, non-prehensile manipulation strategies that leverage transient force closure.