SandWorm: Event-based Visuotactile Perception with Active Vibration for Screw-Actuated Robot in Granular Media

TL;DR

SandWorm couples a bio-inspired rotational–peristaltic locomotion mechanism with SWTac, an actively vibrated, event-based visuotactile sensor, to address perception and locomotion challenges in granular media. The core contributions include hardware-level perception–actuation integration with camera isolation, an IMU-guided temporal filter that improves imaging consistency by about $24$%, and a contact-surface estimation pipeline based on asymmetric edge features and a U‑Net. Empirically, SWTac delivers $0.2$ mm texture resolution, $0.15$ N force estimation accuracy, and $98$% stone-classification accuracy, while SandWorm achieves speeds up to $12.5$ mm/s, successful pipeline dredging, and a $90$% subsurface exploration success in diverse granular media. The work demonstrates that engineered vibration can turn the event camera into a robust static tactile sensor, enabling high-fidelity perception and practical autonomous operation in harsh, contact-rich environments.

Abstract

Perception in granular media remains challenging due to unpredictable particle dynamics. To address this challenge, we present SandWorm, a biomimetic screw-actuated robot augmented by peristaltic motion to enhance locomotion, and SWTac, a novel event-based visuotactile sensor with an actively vibrated elastomer. The event camera is mechanically decoupled from vibrations by a spring isolation mechanism, enabling high-quality tactile imaging of both dynamic and stationary objects. For algorithm design, we propose an IMU-guided temporal filter to enhance imaging consistency, improving MSNR by 24%. Moreover, we systematically optimize SWTac with vibration parameters, event camera settings and elastomer properties. Motivated by asymmetric edge features, we also implement contact surface estimation by U-Net. Experimental validation demonstrates SWTac's 0.2 mm texture resolution, 98% stone classification accuracy, and 0.15 N force estimation error, while SandWorm demonstrates versatile locomotion (up to 12.5 mm/s) in challenging terrains, successfully executes pipeline dredging and subsurface exploration in complex granular media (observed 90% success rate). Field experiments further confirm the system's practical performance.

Figures (24)

• Figure 1: Overview of SandWorm. (a) The snakelike robot is navigating an outdoor trench with the conical elastomer. (b) Manually operated drilling with the frustoconical elastomer on a sandy beach, showing the texture of an ammonoid fossil. (c) Schematic of SandWorm, featuring an integrated active vibration mechanism and hybrid locomotion system. An IMU-guided temporal filter is introduced to enhance imaging quality.
• Figure 2: Overview of SandWorm's mechanical structure. (a) Rendered views of SandWorm: an autonomous snakelike robot and a manually operated device. Cross-sectional views of (b) SWTac, the perception module, (c) the rotational driving module and (d) the snakelike tail module. (e) Exploded view of SWTac.
• Figure 3: Illumination setup and conditions. (a) Hardware setup and optical‑path layout. (b) Reconstructed images of event camera with i. No illumination; ii. Direct dark‐field illumination; iii. Direct bright‐field illumination; iv. Diffuse bright‐field illumination with a diffuser plate.
• Figure 4: Overview of the SandWorm robot's DOFs. Locomotion is achieved by combining rotation, peristaltic translation, and vertical oscillation. The vibrational elastomer tip for perception is also illustrated.
• Figure 5: System architecture of the SandWorm robot, detailing the hardware and algorithm integration. The hardware (bottom) features the SWTac visuotactile sensor (comprising an isolated and a vibrational part) and the locomotive modules. The algorithmic pipeline (top) processes the event stream and IMU readout from the event camera, including an IMU-guided temporal filter followed by downstream tasks.