Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Inverse Design of Snap-Actuated Jumping Robots Powered by Mechanics-Aided Machine Learning

Dezhong Tong, Zhuonan Hao, Mingchao Liu, Weicheng Huang

TL;DR

We address the challenge of designing soft jumping robots with tunable trajectories by combining a DDG-based reduced-order simulator for snap-through actuation with a data-driven inverse design workflow. The approach captures beam buckling, contact, and dynamics, then trains a lightweight forward model to map design parameters to jumping outcomes and uses gradient-based optimization to achieve target jumps. Key contributions include a novel snap-actuated actuator design, an efficient physics-based simulation framework, and a two-stage inverse design method that can plan designs in real time with millimeter accuracy. The work enables rapid design, control, and potential onboard implementation for soft robotic jumping in varied environments.

Abstract

Exploring the design and control strategies of soft robots through simulation is highly attractive due to its cost-effectiveness. Although many existing models (e.g., finite element analysis) are effective for simulating soft robotic dynamics, there remains a need for a general and efficient numerical simulation approach in the soft robotics community. In this paper, we develop a discrete differential geometry-based numerical framework to achieve the model-based inverse design of a novel snap-actuated jumping robot. It is found that the dynamic process of a snapping beam can be either symmetric or asymmetric, such that the trajectory of the jumping robot can be tunable (e.g., horizontal or vertical). By employing this novel mechanism of the bistable beam as the robotic actuator, we next propose a physics-data hybrid inverse design strategy for the snap-jump robot with a broad spectrum of jumping capabilities. We first use the physical engine to study the influences of the robot's design parameters on the jumping capabilities, then generate extensive simulation data to formulate a data-driven inverse design solution. The inverse design solution can rapidly explore the combination of design parameters for achieving a target jump, which provides valuable guidance for the fabrication and control of the jumping robot. The proposed methodology paves the way for exploring the design and control insights of soft robots with the help of simulations.

Inverse Design of Snap-Actuated Jumping Robots Powered by Mechanics-Aided Machine Learning

TL;DR

We address the challenge of designing soft jumping robots with tunable trajectories by combining a DDG-based reduced-order simulator for snap-through actuation with a data-driven inverse design workflow. The approach captures beam buckling, contact, and dynamics, then trains a lightweight forward model to map design parameters to jumping outcomes and uses gradient-based optimization to achieve target jumps. Key contributions include a novel snap-actuated actuator design, an efficient physics-based simulation framework, and a two-stage inverse design method that can plan designs in real time with millimeter accuracy. The work enables rapid design, control, and potential onboard implementation for soft robotic jumping in varied environments.

Abstract

Exploring the design and control strategies of soft robots through simulation is highly attractive due to its cost-effectiveness. Although many existing models (e.g., finite element analysis) are effective for simulating soft robotic dynamics, there remains a need for a general and efficient numerical simulation approach in the soft robotics community. In this paper, we develop a discrete differential geometry-based numerical framework to achieve the model-based inverse design of a novel snap-actuated jumping robot. It is found that the dynamic process of a snapping beam can be either symmetric or asymmetric, such that the trajectory of the jumping robot can be tunable (e.g., horizontal or vertical). By employing this novel mechanism of the bistable beam as the robotic actuator, we next propose a physics-data hybrid inverse design strategy for the snap-jump robot with a broad spectrum of jumping capabilities. We first use the physical engine to study the influences of the robot's design parameters on the jumping capabilities, then generate extensive simulation data to formulate a data-driven inverse design solution. The inverse design solution can rapidly explore the combination of design parameters for achieving a target jump, which provides valuable guidance for the fabrication and control of the jumping robot. The proposed methodology paves the way for exploring the design and control insights of soft robots with the help of simulations.
Paper Structure (16 sections, 15 equations, 6 figures, 1 table)

This paper contains 16 sections, 15 equations, 6 figures, 1 table.

Table of Contents

  1. Introduction
  2. Problem formulation
  3. Numerical Framework
  4. Elastic deformation
  5. Frictional contact force
  6. Time marching scheme
  7. Dynamic Analysis of Jump
  8. Snap-through of elastic beam
  9. Contact-induced jump of robot
  10. Optimal installation position of the actuator
  11. Design parameters of the jump
  12. Inverse design
  13. Data Generation
  14. Learning a forward model
  15. Methodology for the inverse design
  16. ...and 1 more sections

Figures (6)

  • Figure 1: Snapshots for the snap-induced jump robot. (a) Snapshots from experiments. (b) Numerical setup. (c) Snapshots from simulation. The experimental figures are from Ref. wang2023insect.
  • Figure 2: Snap bifurcation of a pre-compressed beam under rotational input. (a1) Geometry of the snap-jump robot. (a2) Discrete elements in numerical simulation. (b1) Equilibrium configurations during the snap-through process. (b2) Static bifurcation diagram. (b3) Energy releases as a function of pre-compression ratio $\varepsilon$. (c1) Dynamic snap-through for asymmetric clamped boundary condition. (c2) Dynamic snap-through for perfect symmetric clamped boundary condition. (c3) Maximum midpoint angle during the snap-through as a function of mismatch angle $\delta \alpha$.
  • Figure 3: (a) Trajectories for jump robot with different design parameters. (b) Maximum jump height and the associated distance, normalized by beam length, as a function of different parameters: (1) beam height, $h$, (2) pre-compression ratio, $\varepsilon$, (3) angle mismatch $\delta \alpha$, (4) frictional coefficient, $\mu$, and (5) normalized mass, $\bar{m}$.
  • Figure 4: Inverse design of the jumping robot. (a) Illustrations of all design parameters, including the desired jump ($x_c^d, y_c^d$), controllable parameters (angle mismatch $\delta\alpha$ and pre-compression ratio $\varepsilon$), and environmental parameters (friction coefficient $\mu$ and normalized mass $\bar{m}$). (b) A lightweight forward model is trained using simulation data to predict the robot's jump based on the given design parameters. (c) The overall inverse design strategy.
  • Figure 5: Influence of environmental parameters on the available region $\mathcal{M}$. (a) Influence of the normalized robot mass $\bar{m}$ with a constant friction coefficient $\mu = 0.3$. (b) Influence of the friction coefficient with a constant robot mass $\bar{m} = 0.768$.
  • ...and 1 more figures