ARCSnake V2: Amphibious Snake Robot

• ARCSnake V2 is an amphibious, screw-propelled snake-like robot designed for versatile locomotion on land, granular, and aquatic terrains.
• It features a water-sealed, hyper-redundant architecture with cable-driven universal joints and Archimedean screw modules enabling screwing, wheeling, and sidewinding modes.
• Experimental tests demonstrate efficient thrust generation, low communication latency, and stable buoyancy control, supporting applications in exploration, search and rescue, and environmental monitoring.

ARCSnake V2 is an amphibious, screw-propelled, snake-like robot engineered for locomotion in multi-domain environments, including land, granular terrains, and aquatic settings. It integrates a hyper-redundant, serially linked, water-sealed mechanical structure with modular screw actuation, cable-driven universal joints, and a distributed buoyancy control system. The system architecture supports real-time teleoperation or autonomous control via a kinematically matched handheld interface and enables transitions among screwing, wheeling, and sidewinding modes. Experimental results demonstrate its performance in thrust, communication latency, buoyancy modulation, and terrain adaptability, positioning ARCSnake V2 as a platform for exploration, search and rescue, and environmental monitoring across heterogeneous domains (Wickenhiser et al., 15 Nov 2025).

1. Mechanical Structure and Actuation

ARCSnake V2 comprises a series of identical, water-sealed segments. Each segment is connected by a 2-DOF cable-driven universal (U-) joint providing ±180° in two orthogonal axes, with each joint capable of 2.6 Nm continuous and 13 Nm peak output torque. Integrated within each segment is an Archimedean-screw propulsion module (maximum 0.269 m diameter, 0.441 m length, 1.0 Nm continuous, 3.8 Nm peak torque) and a positive-pressure plumbing line (operating at 4 psi) that pressurizes the interior to achieve IP67-equivalent water sealing.

The screw drive utilizes a belt-driven motor interfacing a sun gear, planetary stage, and ring gear assembly to rotate the screw shell, while load-bearing is managed by 6656K17 thin-section bearings and cable penetrators with spring-loaded rotary shaft seals for through-segment routing of power, data (CAN), and air lines.

U-joint actuation consists of RMD-X8 Pro motors with compound 5:1 internal and 1.8:1 external reductions, transmitting motion through timing belts with idlers to minimize backlash. A cable coil configuration on the output pulley translates rotational motion into joint bending. The head module augments this with a 6-bar gripper (70 kg-cm IP68 servo), a second 55 kg-cm IP68 servo, and a 135° FOV camera.

A natural Denavit–Hartenberg (D-H) model for each segment, comprising yaw ($\psi_i$), pitch ($\theta_i$), and link length $L_i$, is specified as:

• $$a_{i-1} = 0,\ \alpha_{i-1}=0,\ d_i=L_i,\ \theta_i = \psi_i$$
• $$a_i = 0,\ \alpha_i = \frac{\pi}{2},\ d_{i+1} = 0,\ \theta_{i+1} = \theta_i$$ The resulting homogeneous transformation is provided in explicit matrix form in the source.

2. Screw Propulsion Dynamics

The principal thrust generation in ARCSnake V2 derives from Archimedean screws, leveraging a helical blade to induce axial displacement of the surrounding medium upon rotation. The following parameters define the system:

• $D$: outer diameter of the screw,
• $p$: lead/pitch of the helix,
• $\omega$: rotational speed (rad/s),
• $R = D/2$,
• $\alpha = \arctan\left(\frac{p}{\pi D}\right)$, the helix angle.

Governing relations include:

• Tip speed: $v_{tip} = \omega R$
• Slip ratio: $s = 1 - \frac{v_{medium}}{v_{tip}}$
• Axial thrust: $F_{thrust} = T \sin\alpha$, where $T = \eta \tau_{motor}$ (with $\eta$ denoting reduction efficiency)
• Radial force: $F_{rad} = T \cos\alpha$

Experimental characterization with $\omega$ from 10 to 50 rad/s yielded thrusts from 40.0 N to 75.9 N. Drive efficiency was calculated as approximately $65.7\%$ based on measured no-load and loaded torque values. In granular media, soil reaction forces further influence effective thrust, but validation used empirical measurement of net tangential loads.

3. Buoyancy and Pressure Management

Each segment is designed with a dual-source buoyancy system:

• Passive: Internal filling with polyurethane marine foam.
• Active: Inflatable torus-shaped bladder encircling each U-joint.

Archimedes’s law governs net buoyant force: $F_b = \rho_w g V_{disp}$, where $\rho_w = 1000$ kg/m³, $g = 9.81$ m/s², and $V_{disp}$ is volume displaced.

The torus bladder volume is: $V_b = 2\pi^2 r^2 R$, with $r$ and $R$ representing minor and major radii. A $\Delta F_b \approx 10$ N shift (corresponding to a change between –5% and +5% overall buoyancy) requires $V_b \approx 1.43 \times 10^{-3}$ m³ per bladder.

The pneumatic system operates as follows:

• Bladders fill at $\approx$ 70 s (2.9 psia); upstream regulators are set to 3.0 psia.
• A positive-pressure (6 psi) line prevents water ingress.

Buoyancy actuation is currently teleoperated via manual valve control, although a proportional control law can be expressed as $\Delta V = k_p(z_{cmd} - z)$, relating commanded and actual depth.

4. Locomotion Modes and Gait Transitions

ARCSnake V2 enables multiple locomotion strategies with smooth mode switching mediated by joint and actuator command ramping, informed by internal IMU and depth sensors.

Screwing: All U-joints are aligned; helical screws generate propulsion. Three-dimensional curvilinear motion is parameterized by projecting measured head velocity ($\mathbf{v}_h$) and angular velocity ($\boldsymbol{\omega}_h$) onto the local motion plane, then calculating turning radius: $$R = \frac{||\mathbf{v}_\perp||}{||\boldsymbol{\omega}_\perp||}$$, with a constant curvature joint configuration $$\theta = 2 \tan^{-1}\left(\frac{l ||\boldsymbol{\omega}_\perp||}{||\mathbf{v}_\perp||}\right)$$.

Wheeling: U-joints are configured into a body-spanning loop (hoop), using circumferential screws as wheels. Kinematic constraints ensure each screw axis is tangent to the defined motion circle.

Sidewinding: Implemented via phase-shifted, alternating vertical and horizontal joint actuation. The joint command laws are: $\psi_i(t) = A_y \sin(2\pi f t + \phi_i)$, $\theta_i(t) = A_p \cos(2\pi f t + \phi_i + \delta)$, where $A_{y/p}$, $f$, $\phi_i$, and $\delta$ define gait amplitude and phase properties. This approach derives from established CPG-based sidewinding strategies in prior snake robotics literature.

Transition between modes is performed incrementally, ensuring uninterrupted stability and control.

5. Electronics, Control, and Teleoperation

The electrical backbone consists of a 48 V tethered main bus, with on-segment PCBs (Vicor V48B24C250BL) providing 24 V to motors and controllers. Communication employs a CAN-bus daisy-chain internal to the snake, connected via a Fathom ROV tether to a host PC. CAN node latency is 0.73 ms (single node), with an incremental 0.91 ms per additional node—yielding a maximum of ~8.92 ms at nine segments.

Orientation and joint angles are measured by onboard IMUs via I²C. The handheld controller offers kinematically matched teleoperation; a 6-DOF spatial joystick maps directly to head pose, and a thumb wheel sets screw speed ($\omega_{cmd}$). Joint commands are issued as $q_i = K_{ij} \eta_j$, where $K$ indexes the contribution of user input to individual joints. Teleoperated bladder inflation/deflation is actuated by two pneumatic solenoids (front/rear).

6. Experimental Performance and Validation

Performance validation covers mechanical, control, and environmental interaction domains:

• Screw Drive: At $\omega_{cmd} = 10$ rad/s, tangential force $F_{tan} = 40.0$ N and shell torque $\tau_{shell} = 3.60$ Nm; at $\omega_{cmd} = 50$ rad/s, $F_{tan} = 75.9$ N and $\tau_{shell} = 6.83$ Nm. Measured drive efficiency is approximately 65.7%.
• Communication Latency: CAN round-trip time is 0.73 ms (base) plus 0.91 ms per segment, yielding sub-10 ms response even at maximal system length.
• Buoyancy System: Displacement volumes are empirically verified. Rear bladder inflates in 68 s, front in 70 s at 2.9 psia. Sinking and rising accelerations are 0.045 m/s² and 0.027 m/s², respectively, under specified pressure.
• Joint Hysteresis: Hysteresis under no load is approximately 2°; with 6.25 lb and 11.25 lb applied loads, hysteresis increases to 4° and 6°, respectively.
• Terrain Mobility: The robot traversed grass, concrete, mulch, and gravel in both screwing and wheeling modes. Maximum achieved speed is $\approx 0.1$ m/s on rigid surfaces, $\approx 0.05$ m/s in granular settings.
• Aquatic Maneuvers: ARCSnake V2 maintained neutral buoyancy for sustained straight swimming and completed teleoperated navigation tasks, including object retrieval with the integrated gripper.

7. Applications and Research Context

ARCSnake V2 advances robotic locomotion strategies for challenging environments where conventional wheeled or legged platforms are limited by uneven, compliant, or submerged substrates. Its multi-domain capability, water-sealed modularity, and force-regulated actuation facilitate deployment for exploration, search and rescue, and environmental monitoring. The combination of distributed electronics, robust CAN-based control, and kinematically mapped teleoperation delivers low-latency interaction and high mobility. The integration of screw propulsion, buoyancy modulation, and hyper-redundant articulated structure embodies a convergence of classical snake robotics and modern amphibious systems (Wickenhiser et al., 15 Nov 2025).

A plausible implication is the system’s adaptability as a prototyping platform for future autonomous multi-terrain robots or as a research tool for studies in bio-inspired locomotion with in-situ environmental interaction.