Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

GeoDEx: A Unified Geometric Framework for Tactile Dexterous and Extrinsic Manipulation under Force Uncertainty

Sirui Chen, Sergio Aguilera Marinovic, Soshi Iba, Rana Soltani Zarrin

TL;DR

GeoDEx addresses tactile manipulation under force-uncertainty by unifying force estimation, planning, and control within a geometric framework. It leverages representations like the FE-plane, FE-basis, and M-cone, together with a trusted-ellipsoid bound, to robustly plan and estimate contact forces and drive an admittance-based controller. Hardware experiments across 2-, 3-, and 4-finger grasps and extrinsic manipulation demonstrate substantially improved force tracking and grasp success compared with raw tactile readings, complemented by a ~14× speed-up over SOCP-based planning in simulation. This approach reduces reliance on highly calibrated force sensors and enables reliable dexterous and extrinsic manipulation in the presence of force-reading noise.

Abstract

Sense of touch that allows robots to detect contact and measure interaction forces enables them to perform challenging tasks such as grasping fragile objects or using tools. Tactile sensors in theory can equip the robots with such capabilities. However, accuracy of the measured forces is not on a par with those of the force sensors due to the potential calibration challenges and noise. This has limited the values these sensors can offer in manipulation applications that require force control. In this paper, we introduce GeoDEx, a unified estimation, planning, and control framework using geometric primitives such as plane, cone and ellipsoid, which enables dexterous as well as extrinsic manipulation in the presence of uncertain force readings. Through various experimental results, we show that while relying on direct inaccurate and noisy force readings from tactile sensors results in unstable or failed manipulation, our method enables successful grasping and extrinsic manipulation of different objects. Additionally, compared to directly running optimization using SOCP (Second Order Cone Programming), planning and force estimation using our framework achieves a 14x speed-up.

GeoDEx: A Unified Geometric Framework for Tactile Dexterous and Extrinsic Manipulation under Force Uncertainty

TL;DR

GeoDEx addresses tactile manipulation under force-uncertainty by unifying force estimation, planning, and control within a geometric framework. It leverages representations like the FE-plane, FE-basis, and M-cone, together with a trusted-ellipsoid bound, to robustly plan and estimate contact forces and drive an admittance-based controller. Hardware experiments across 2-, 3-, and 4-finger grasps and extrinsic manipulation demonstrate substantially improved force tracking and grasp success compared with raw tactile readings, complemented by a ~14× speed-up over SOCP-based planning in simulation. This approach reduces reliance on highly calibrated force sensors and enables reliable dexterous and extrinsic manipulation in the presence of force-reading noise.

Abstract

Sense of touch that allows robots to detect contact and measure interaction forces enables them to perform challenging tasks such as grasping fragile objects or using tools. Tactile sensors in theory can equip the robots with such capabilities. However, accuracy of the measured forces is not on a par with those of the force sensors due to the potential calibration challenges and noise. This has limited the values these sensors can offer in manipulation applications that require force control. In this paper, we introduce GeoDEx, a unified estimation, planning, and control framework using geometric primitives such as plane, cone and ellipsoid, which enables dexterous as well as extrinsic manipulation in the presence of uncertain force readings. Through various experimental results, we show that while relying on direct inaccurate and noisy force readings from tactile sensors results in unstable or failed manipulation, our method enables successful grasping and extrinsic manipulation of different objects. Additionally, compared to directly running optimization using SOCP (Second Order Cone Programming), planning and force estimation using our framework achieves a 14x speed-up.
Paper Structure (20 sections, 3 theorems, 17 equations, 17 figures, 3 tables)

This paper contains 20 sections, 3 theorems, 17 equations, 17 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related work
  3. State of Tactile Sensors
  4. Utilizing Tactile Readings
  5. Tactile-Enabled Dexterous Manipulation
  6. Formulation and Methodology
  7. Definitions
  8. Force Estimation
  9. Force Planning
  10. Controller
  11. Experiments and Results
  12. Force Measurement via Tactile Array
  13. Simulation Results
  14. Hardware Results
  15. 2-finger grasp
  16. ...and 5 more sections

Key Result

Proposition 1

On the FE-plane, we can construct a set of orthonormal vectors that define a basis $\bm{B}_{fe}=\{\bm{b}_1,...,\bm{b}_{d_{fe}}\}, \bm{b}_i\in\mathbb{R}^{(n_i+n_e)\times3}$. Using this FE-basis, we can represent any force vector on the FE-plane using a weighted sum of these basis vectors $\bm{f}_{fe}

Figures (17)

  • Figure 1: System diagram of our proposed method
  • Figure 2: FE-plane, M-Cone and Constraint convex set
  • Figure 3: Illustration of measurement sub-space cone Assume two contact points are involved, one intrinsic, one extrinsic. Orange ray shows measurement sub-cone as we can only measure force magnitude at the intrinsic contact location.
  • Figure 4: Block diagram of the control system.
  • Figure 5: Hardware setup including Allegro hand equipped with Touchlab fingertips, and Franka arm. The $3D$-printed sphere and wrench, as well as the cylindrical can are used for dexterous grasping experiments, while the $3D$-printed cube and a real screwdriver are used for extrinsic manipulation experiments.
  • ...and 12 more figures

Theorems & Definitions (9)

  • Definition 1: Space of contact forces
  • Definition 2: FE-plane
  • Proposition 1: FE-basis
  • Definition 3: Constraint convex set
  • Proposition 2: Perservation of force equilibrium
  • proof
  • Definition 4: M-Cone
  • Definition 5: Trusted measurement ellipsoid
  • Proposition 3