[Paper Review] Four-Arm Manipulation via Feet Interfaces

Jacob Hernandez Sanchez, Walid Amanhoud+4 more|arXiv (Cornell University)|Sep 11, 2019

Muscle activation and electromyography studies24 references9 citations

TL;DR

This paper presents a novel four-arm manipulation system where humans control two robotic arms using their feet via custom 5-DOF foot interfaces, enabling dexterous hand use while feet handle support tasks. Using impedance-controlled teleoperation with dynamical systems for trajectory generation and haptic feedback, the system achieves high force transparency (RMSE 0.033 N) and effective disturbance rejection during contact, validating direct foot-based control for supernumerary robotic limbs.

ABSTRACT

We seek to augment human manipulation by enabling humans to control two robotic arms in addition to their natural arms using their feet. Thereby, the hands are free to perform tasks of high dexterity, while the feet-controlled arms perform tasks requiring lower dexterity, such as supporting a load. The robotic arms are tele-operated through two foot interfaces that transmit translation and rotation to the end effector of the manipulator. Haptic feedback is provided for the human to perceive contact and change in load and to adapt the feet pressure accordingly. Existing foot interfaces have been used primarily for a single foot control and are limited in range of motion and number of degrees of freedom they can control. This paper presents foot-interfaces specifically made for bipedal control, with a workspace suitable for two feet operation and in five degrees of freedom each. This paper also presents a position-force teleoperation controller based on Impedance Control modulated through Dynamical Systems for trajectory generation. Finally, an initial validation of the platform is presented, whereby a user grasps an object with both feet and generates various disturbances while the object is supported by the feet.

Motivation & Objective

• To develop a mechanical and control system enabling two robotic arms to be teleoperated via both feet simultaneously, extending human manipulation capacity.
• To address the challenge of controlling robotic arms with feet, which must handle the leg's weight and provide sufficient dexterity and workspace for coordinated manipulation.
• To enable hands-free operation where natural arms perform high-dexterity tasks while feet-controlled arms manage support or force-intensive tasks.
• To validate the feasibility of direct foot-based teleoperation with haptic feedback for stable, transparent, and responsive control in contact scenarios.

Proposed method

• Design of two 5-DOF foot interfaces using serial kinematics to maximize workspace and allow close proximity operation of both feet.
• Implementation of a position-force teleoperation controller based on impedance control modulated by dynamical systems for smooth trajectory generation.
• Use of Denavit-Hartenberg (DH) kinematic modeling with supplementary frames to represent Tait-Bryan angles and avoid singularities.
• Incorporation of haptic feedback through force sensors in the foot platforms, reflecting interaction forces to the user for real-time adaptation.
• Employment of low gear ratio and backdrivable joints to ensure high transparency and low force error in force reflection.
• Use of attractor dynamics in the dynamical system to guide motion convergence and improve compliance during contact tasks.

Experimental results

Research questions

• RQ1 Can a dual-foot interface with five degrees of freedom per foot enable stable, coordinated control of two robotic arms for manipulation tasks?
• RQ2 How effective is haptic feedback in enabling users to perceive contact forces and adapt pressure during object grasping and disturbance application?
• RQ3 To what extent does impedance control with dynamical system modulation improve motion smoothness and force tracking in foot-based teleoperation?
• RQ4 What is the performance of the system in terms of position and force tracking accuracy during free motion and contact phases?
• RQ5 Can the system maintain stability and transparency under abrupt external disturbances, such as hammering?

Key findings

• The system achieved high force transparency in the task-aligned direction, with a root mean squared error (RMSE) of only 0.033 N during contact phases.
• Position tracking error was lower in the direction of grasping (RMSE_y = 0.087 m) during free motion compared to contact phases (RMSE_y = 0.107 m), likely due to virtual attractor pushing into constrained object geometry.
• In orthogonal directions, tracking error was lower during contact (RMSE_x = 0.050 m, RMSE_z = 0.060 m) than in free motion (RMSE_x = 0.132 m, RMSE_z = 0.074 m), indicating better stability and convergence under constraints.
• The impedance control strategy effectively smoothed motion, reducing jerkiness and enhancing stability, particularly during abrupt disturbances like hammering.
• The system successfully maintained a bipedal grasp under external perturbations, demonstrating robustness and user adaptability.
• The open-loop haptic control with approximate dynamic compensation yielded small force errors, suggesting that closed-loop force control may not be necessary for initial implementations.