MIGHTY Suction Cup: Sensorized Robotic Gripper

• MIGHTY Suction Cup is a sensorized suction gripper integrating force/torque sensing to provide real-time feedback for safe robotic manipulation.
• It employs adaptive admittance control with variable damping to reduce operator effort and enhance human–robot collaboration.
• A barrier artificial potential mechanism ensures grip safety by proactively preventing detachment through sensorless corrective torque.

The MIGHTY Suction Cup encompasses a set of compliant, sensor-integrated suction gripper technologies designed for advanced robotic manipulation, collaborative transport, and state estimation in unstructured and dynamic environments. The distinctive feature of the MIGHTY Suction Cup is its dual role: serving both as a secure mechanical gripper and as a force/torque sensor, thereby underpinning high-dexterity human–robot interaction, robust object handling, and safe, adaptive manipulation. In state-of-the-art implementations, the MIGHTY Suction Cup has been employed as the end-effector of legged robots—such as quadrupeds engaged in collaborative object transport—where it enables compliant physical interaction and incorporates formal control-theoretic guarantees on safety and passivity (Plotas et al., 1 Aug 2025).

1. Suction Cup as Force/Torque Sensor and Gripper

The MIGHTY Suction Cup is physically constructed as a conventional suction gripper retrofitted with embedded multi-chamber force/torque sensing elements. The suction cup assembly typically comprises:

• A flexible cup fabricated from elastomeric material (e.g., silicone rubber), forming a seal on the object’s surface when negative pressure is applied.
• Four sensorized internal chambers arranged symmetrically within the cup. Each chamber is plumbed to an external high-precision pressure transducer via vacuum tubing.
• The negative pressure acts both to create adhesion (gripping force) and to provide independent pressure measurement signals from each chamber.

This construction enables the suction cup to measure the distribution of holding forces at the interface, thereby acting as a full six-dimensional force/torque sensor when the transformation from chamber readings to resultant force/torque is known. In the context of a manipulator or legged robot, these measurements are mapped via a rigid transformation to the body frame, allowing direct feedback for interaction control (Plotas et al., 1 Aug 2025).

2. Admittance Control and Variable Damping for Human-Robot Collaboration

In collaborative transportation, the MIGHTY Suction Cup underpins an admittance control framework that yields compliant, human-driven robot motion. The admittance controller imposes a virtual mass-damper second-order differential relationship:

$M_d\,\dot{v}_b + D_d\,v_b = \Lambda F_b + F_v$

where $v_b$ is the reference body velocity (planar and yaw), $M_d$ is a user-defined inertia matrix, $D_d$ is the damping matrix, $F_b$ is the mapped force (from the suction cup sensor), $\Lambda$ a task-space selection matrix, and $F_v$ is an auxiliary command to prevent detachment.

A key innovation is the variable damping $D_d(P^+)$, where the scalar damping factor $\zeta(P^+)$ adapts to the instantaneous power input $P^+ = \max(0, v_b^\top\Lambda F_b)$. The functional form is:

$$\zeta(P^+) = \zeta_\ell + (\bar{\zeta} - \zeta_\ell)\exp(-\lambda P^+)$$

which decreases damping (promoting motion) when the human is actively moving the object, and increases damping (providing resistance and stability) when the human intends to stop or slow the system. This enables both increased controllability and reduced operator effort (Plotas et al., 1 Aug 2025).

3. Barrier Artificial Potential for Grip Safety

Object detachment risk is addressed through a barrier artificial potential (BAP) designed to ensure the minimum measured suction force in any chamber, $\min(f_s)$, remains above a critical threshold $f_{\text{min}}$. The BAP is constructed as:

$$W = \begin{cases} \frac{k_1}{2}\left(\frac{1}{f_m} - \frac{1}{f_0}\right)^2 + \frac{k_2}{2}(f_m - f_0)^2, & \text{if } f_m < f_0 \ 0, & \text{otherwise} \end{cases}$$

where $f_m$ is a smooth (exponentially weighted) minimum over chamber forces, $f_0$ is a safety margin, and $k_1$, $k_2$ are gains. The added control signal $F_v = [0, 0, -\partial W/\partial\theta]^\top$ acts as a virtual torque about the yaw axis, steering the robot configuration away from imminent loss of grip.

This mechanism is fully sensorless in actuation, relying exclusively on multi-point pressure measurements provided by the MIGHTY Suction Cup. The design ensures that as the grip approaches the detachment threshold, the BAP grows rapidly, triggering corrective behaviors before actual loss (Plotas et al., 1 Aug 2025).

4. Passivity Analysis and Theoretical Guarantees

A distinguishing aspect of the control architecture is the rigorous guarantee of passivity. The storage function

$\mathcal{L} = \frac{1}{2}v_b^\top M_d v_b + W$

serves as a Lyapunov function, with $$\dot{\mathcal{L}} \leq v_b^\top u - \text{dissipative terms}$$ along system trajectories, where $u$ is the input of human-applied force. Hence, the system always dissipates or stores energy provided by the operator, never generating its own, ensuring safety in collaborative scenarios regardless of model uncertainties or external disturbances.

This mathematical structure guarantees that, under continuous or impulsive contact, the system cannot amplify energy—a property essential in physical human–robot interaction, especially when handling heavy or fragile objects collaboratively (Plotas et al., 1 Aug 2025).

5. Experimental Realization on Quadruped Platforms

The MIGHTY Suction Cup has been empirically validated on the Unitree Go1 quadruped platform for collaborative tasks. Specific experiment types include:

• Arc Motion (Rotation): The robot, while carrying an object via the suction cup, is guided in a 90° arc by a human. When the BAP is active, the controller senses dropping chamber forces and introduces corrective torques to preempt detachment. Without BAP, detachment is observed near 5.1 s.
• Translation: The robot is led forward while damping strategies are varied. Variable damping shows a ~2.6× reduction in operator energy requirement relative to a constant high-damping baseline, and eliminates excessive oscillations characteristic of low damping.

Figures and time-series plots demonstrate the real-time evolution of chamber forces, BAP response, commanded velocities, and cumulative operator energy, all supporting improved safety-effort tradeoff and grip assurance (Plotas et al., 1 Aug 2025).

6. System Architecture and Signal Pathway

A conceptual signal flow underlying the MIGHTY Suction Cup in collaborative transport is:

[Force Sensor (MIGHTY)]–(measure forces fᵢ)→[Transform to robot body F_b]→ [Admittance Controller (M_d, D_d)]→[Variable Damping ζ(P⁺)]→[BAP Safety Term F_v]→[Task-space velocity reference v_b]→[Robot motion execution]

This encapsulates the closed-loop control chain, fully utilizing the dual sensing and gripping functions of the suction end-effector.

7. Impact, Limitations, and Extensions

The deployment of the MIGHTY Suction Cup establishes a power- and safety-aware bridge between human operators and articulated robotic systems. The tension between high compliance (easy guiding) and grip safety (detachment prevention) is resolved through adaptive damping and barrier potential mechanisms founded on sensorized suction hardware.

A recognized limitation is the reliance on accurate calibration of chamber force measurements and correct mapping of these to the control policy for arbitrary object geometries and cup mounting positions. Closed-loop performance may depend on the fidelity of the force-to-torque Jacobian and on well-tuned controller parameters (e.g., $f_{\min}$, $k_1$, $k_2$). A plausible implication is the need for surface- or payload-specific calibration procedures to maximize robustness.

The approach generalizes to other physical human–robot interaction use-cases where safe, energy-bounded compliance and grip assurance are needed. Combining the MIGHTY Suction Cup’s feedback with vision or external tactile sensing could further improve versatility in complex settings.

---

In summary, the MIGHTY Suction Cup, as detailed in (Plotas et al., 1 Aug 2025), is a compliant, sensorized gripper that enables simultaneous physical interaction sensing and safe object manipulation. Its architecture supports passivity-guaranteed admittance control and barrier potential grip maintenance, validated experimentally on legged robots in collaborative transport, and is a representative example of integrative hardware-software co-design in modern human–robot collaboration.