cs.ROMar 8, 2026

A Robust Antenna Provides Tactile Feedback in a Multi-legged Robot

Authors: Zhaochen J. Xu, Juntao He, Delfin Aydan, Malaika Taylor, Tianyu Wang, Jianfeng Lin, Wesley Dyer, Daniel I. Goldman

Organizations: All authors are with the Georgia Institute of Technology, Atlanta, GA 30332, USA.

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Multi-legged elongate robots hold promise for maneuvering through complex environments. Prior work has demonstrated that reliable locomotion can be achieved using open-loop body undulation and foot placement on rugose terrain. However, robust navigation through confined spaces remains challenging when body-environment contact is extensive and terrain rheology varies rapidly. To address this challenge, we develop a pair of tactile antennae for multi-legged robots that enable real-time sensing of surrounding geometry, modeling the morphology and function of biological centipede antennae. Each antenna features gradient compliance, with a stiff base and soft tip, allowing repeated deformation and elastic recovery. Robophysical experiments reveal a relationship between continuous antenna curvature and contact force, leading to a simplified mapping from antenna deformation to inferred discrete collision states. We incorporate this mapping into a controller that selects among a set of locomotor maneuvers based on the inferred collision state. Experiments in obstacle-rich and confined environments demonstrate that tactile feedback enables reliable steering and allows the robot to recover from near-stuck conditions without requiring global environmental information or real-time vision. These results highlight how mechanically tuned tactile appendages can simplify sensing and enhance autonomy in elongate multi-legged robots operating in constrained spaces.

Figures & tables

Fig. 1: Robophysical antenna model integrated on a multi-legged robot (top-down view). (A) Antenna design; The antenna enables compliant contact sensing through measured bending. (B) Top-down view of the centipede S. subspinipes showing wall-sensing of antenna. (C) Top-down view of the multi-legged robot SCUTTLE (Ground Control Robotics, Inc.). The dashed box highlights the modular antenna mounted on the anterior segment.
Fig. 2: Antenna morphology of a centipede. (A) Lateral view showing the cephalic plate, antennae, ocelli, and forcipules, with anterior direction and sagittal plane indicated (scale bar: 1 cm). (B) Anterior view highlighting the bilaterally extended antennae and their descending segment diameter from base to tip (proximal diameter 0.15 cm; distal diameter 0.02 cm). (C) Diameter based stiffness proxy along the antenna, computed from the proximal to distal diameter profile using a fourth power scaling ( K^i∝d4\hat{K}_{i}\propto d^{4} ), showing a descending stiffness distribution from base to tip.
Fig. 3: Antenna contact and in-contact antenna morphologies during wall-following in S. subspinipes (top-down view). (A) Representative time sequence showing sustained antenna contact with a wall during forward travel (scale bar: 2 cm). (B) Three stereotyped in-contact antenna morphologies, defined by the direction the antenna points relative to the body: forward pointing , backward pointing , and straight . (C) Distribution of in-contact antenna morphologies across trials ( n=4n=4 ), including a transition category for brief switching between morphologies during continuous contact.
Fig. 4: Robotic antenna modes and actuation schematic. (A) Representative in-contact antenna morphologies during wall interaction, labeled by the direction the antenna points relative to the body ( forward pointing and backward pointing ). (B) Planar antenna sweep used for calibration and testing, parameterized by a sinusoidal base angle θjoint(t)=θ0+Asin⁡(2πft)\theta_{\mathrm{joint}}(t)=\theta_{0}+A\sin(2\pi ft) .
Active Coils ( NcN_{c} )ktorsionk_{\mathrm{torsion}} (N ⋅\cdot mm/deg)Position
30.274Base (Stiffest)
60.137Middle (Moderate)
90.091Tip (Compliant)
TABLE I: Theoretical Torsional Stiffness by Active Coils
Fig. 5: Antenna calibration and mapping from ADC to normalized bending level. Raw 12-bit ADC readings are recorded while the antenna is swept through controlled deflections at three traverse speeds ( v=5,10,15v=5,10,15 cm/s). Markers show the mean and error bars indicate variability across trials. A global decreasing sigmoid fit (inset) defines the mapping from averaged ADC to a normalized bending level b∈[0,1]b\in[0,1] used by the controller. Fit parameters for this dataset: ks=18.3k_{s}=18.3 , x0=0.305x_{0}=0.305 .
Fig. 6: Effect of antenna compliance on contact interaction. (A) Interchangeable torsion springs set three stiffness conditions (stiff: Nc=3N_{c}{=}3 , ktorsion=0.274k_{\mathrm{torsion}}{=}0.274 N ⋅\cdot mm/deg; medium: Nc=6N_{c}{=}6 , ktorsion=0.137k_{\mathrm{torsion}}{=}0.137 N ⋅\cdot mm/deg; compliant: Nc=9N_{c}{=}9 , ktorsion=0.091k_{\mathrm{torsion}}{=}0.091 N ⋅\cdot mm/deg). Too rigid (B) increases contact forces and causes jamming and too compliant (D) leads to collapse and missed contact; the smooth descending stiffness profile (C) is optimal with high temporal resolution reading without getting stuck. (E) Bar chart summarizes traversal success rate for the three stiffness profile.
Fig. 7: Flowchart of the discrete antenna-based controller for confined traversal. Bilateral antenna signals are mapped through a sigmoid to contact variables, which drive the decision process with a time window holding. The controller selects among forward, turn, and reverse maneuvers based on contact presence, symmetry, and dominance to provide close-loop navigation in confined space.
Fig. 8: Experiment on robotic platform SCUTTLE. (A) Modular multi-legged robot with enclosed cabin, eccentric O-legs, body joint servos, and a bilateral antenna module (top-down view inset showing traveling-wave gait). (B) Overhead view of cluttered traversal showing representative trajectory with close-loop control.
More in this topic

Legged Robotics

Legged robotics studies how robots with legs move, perceive their surroundings, and perform useful tasks across varied terrain.

See all 126 papers

Explore similar work

CardsList
Feb 20, 2026

Soft Surfaced Vision-Based Tactile Sensing for Bipedal Robot Applications

Legged locomotion benefits from embodied sensing, where perception emerges from the physical interaction between body and environment. We present a soft-surfaced, vision-based tactile foot sensor that endows a bipedal robot with a skin-like deformable layer that captures contact deformations optically, turning foot-ground interactions into rich haptic signals. From a contact image stream, our method estimates contact pose (position and orientation), visualizes shear, computes center of pressure (CoP), classifies terrain, and detects geometric features of the contact patch. We validate these capabilities on a tilting platform and in visually obscured conditions, showing that foot-borne tactile feedback improves balance control and terrain awareness beyond proprioception alone. These findings suggest that integrating tactile perception into legged robot feet improves stability, adaptability, and environmental awareness, offering a promising direction toward more compliant and intelligent locomotion systems. For the supplementary video, please visit: https://youtu.be/ceJiy9q_2Aw

Jaeeun Kim, Junhee Lim, Yu She

Department of Industrial Engineering, Purdue University, West Lafayette, IN 47906, USA · Department of Computer Science, Purdue University, West Lafayette, IN 47906, USA

Robotic Tactile SensingVision-Based Tactile SensingLegged RoboticsRobot LocomotionTerrain-Aware Robot Locomotion
AskPDF
Mar 10, 2026

Walking on Rough Terrain with Any Number of Legs

Robotics would gain by replicating the remarkable agility of arthropods in navigating complex environments. Here we consider the control of multi-legged systems which have 6 or more legs. Current multi-legged control strategies in robots include large black-box machine learning models, Central Pattern Generator (CPG) networks, and open-loop feed-forward control with stability arising from mechanics. Here we present a multi-legged control architecture for rough terrain using a segmental robot with 3 actuators for every 2 legs, which we validated in simulation for robots with 6 to 16 legs. Segments have identical state machines, and each segment also receives input from the segment in front of it. Our design bridges the gap between WalkNet-like event cascade controllers and CPG-based controllers: it tightly couples to the ground when contact is present, but produces fictive locomotion when ground contact is missing. The approach may be useful as an adaptive and computationally lightweight controller for multi-legged robots, and as a baseline capability for scaffolding the learning of machine learning controllers.

Zhuoyang Chen, Xinyuan Wang, Shai Revzen

The authors are with the University of Michigan, Ann Arbor, MI 48109, USA.

Legged RoboticsRobot LocomotionTerrain-Aware Robot LocomotionDynamical Systems
AskPDF
Jul 11, 2026

TAC-LOCO: Unified Whole-Body Control for Quadrupedal TACtile-Informed LOCO-Manipulation

Dynamic loco-manipulation requires legged robots to coordinate whole-body motion while maintaining stable physical interaction with grasped objects under uncertain external forces. While tactile sensing has been widely studied for robotic manipulation, its role in dynamic whole-body control remains largely unexplored. Existing works without tactile feedback commonly grasp firmly rather than regulate the grasp according to the interaction. We propose TAC-LOCO, a tactile-augmented unified reinforcement learning framework that encodes tactile array observations from compliant grippers into a compact latent representation and joins it with proprioception for unified control of the legs, arm, and gripper. With effective grasp stability reward design, the policy learns to simultaneously track body velocity and end-effector trajectories, moderate grasp force, and prevent object slip under both gradual load changes and sudden release events. We deploy the policy zero-shot on a Unitree Go2 with an Interbotix WidowX 250 arm and tactile gripper, demonstrating dynamic tactile-informed loco-manipulation under varying external interactions, achieving a 47% reduction in grasping force and an object drop rate of less than 1%.

Muqun Hu, Yuhao Zhou, Kabir Ray Malik +4

1Purdue University · University of Florida

Legged RoboticsRL for RoboticsWhole-Body ControlTactile Sensing for Robotic Manipulation
AskPDF