Object Manipulation of the Variable Topology Truss system

Technical Summary: Object Manipulation of the Variable Topology Truss (VTT) System

Problem Statement
While truss robot systems have been studied for their versatility in rapid deployment and mobile exploration, their capabilities for object manipulation remain underdeveloped. Although kinematic analysis and trajectory planning have been addressed in prior literature, specific strategies for task-space force control and concurrent position-force regulation have not been thoroughly investigated. The Variable Topology Truss (VTT) system, designed for unstructured environments like disaster zones, faces unique challenges in manipulation: the high friction inherent in its spiral zipper actuators and the complex kinematics and statics of the truss structure make it difficult to develop controllers that can accurately track both contact point position and force. Furthermore, existing studies have largely relied on simulations or single-actuator experiments, lacking experimental validation of force control in full truss configurations.

Methodology
To address these challenges, the authors propose a hierarchical hybrid control framework that regulates position and force concurrently without explicit decoupling.

• Hardware and Actuation: The VTT system utilizes member modules equipped with spiral zipper actuators, which offer large extension ratios but suffer from high internal friction. Each module includes a loadcell for force sensing, an ESP32 microcontroller for low-level control, and wireless communication capabilities.
• Control Architecture:
  • Low-Level Control: To overcome actuator friction, a sensor-based force feedback controller (Proportional-Integral) is implemented at the actuator level. This controller regulates the motor's PWM input to generate desired axial forces based on loadcell measurements.
  • High-Level Control: A static model of the VTT system is used to map desired task-space forces at target nodes to the required internal member forces. The inverse Jacobian transpose is utilized to compute the necessary member forces ($\lambda$) to achieve a desired nodal force ($F_{des}$), accounting for external forces and gravity.
  • Hybrid Strategy: The core innovation is a unified control law that generates a force command ($F_{cmd}$) by summing the desired force and a proportional feedback term based on position error ($P_{des} - P_{curr}$). This approach avoids the traditional decoupling of force and position into orthogonal subspaces, allowing the system to control both variables along the same direction simultaneously.
• Experimental Configurations: The study evaluates the system using two configurations:
  1. Double Tetrahedron: A fully constrained structure used to test combined position and force tracking across linear and circular trajectories.
  2. Octahedron: An over-constrained structure used to demonstrate a practical manipulation task involving the transport and stacking of objects.

Key Results

• Actuator-Level Performance: Single-module tests confirmed that the spiral zipper actuators could reliably generate and maintain axial forces with less than 5% error during force-holding phases, despite a "dead zone" at low forces caused by internal friction.
• System-Level Force Control: Experiments on both tetrahedral and pyramid (over-constrained) configurations demonstrated that the system could track target forces within approximately 5% error during holding phases. While overshoot was observed at lower forces due to the actuator dead zone and joint friction, the control method proved effective for both fully and over-constrained structures.
• Manipulation Performance:
  • In the double-tetrahedron configuration, the system successfully tracked linear and circular trajectories while maintaining a grasping force of 60 N. Position Root Mean Square Error (RMSE) remained within centimeters. The study noted a trade-off where position accuracy was prioritized over force tracking precision due to the non-decoupled control gains.
  • In the octahedron configuration, the system successfully executed a complex task: grasping a 309 g box, transporting it 0.5 m, placing it down, and repeating the process with a second box. The entire sequence took approximately 4 minutes and 50 seconds, demonstrating repeatable manipulation.

Significance and Claims
The paper claims to present the first experimental demonstration of object manipulation using a VTT system with a hierarchical hybrid control framework. The authors assert that their approach enables consistent and reliable manipulation in unstructured environments by solving the difficulty of tracking contact point position and force simultaneously. By validating the system on both fully and over-constrained configurations, the study establishes a foundation for autonomous manipulation in disaster response scenarios, such as delivering medical supplies or removing debris. The work confirms that despite the high friction of spiral zipper actuators and complex statics, a sensor-based hybrid control strategy can achieve the necessary precision for practical object manipulation tasks.