Function based sim-to-real learning for shape control of deformable free-form surfaces

Yingjun Tian,Guoxin Fang,Renbo Su,Weiming Wang,Simeon Gill,Andrew Weightman,Charlie C.L. Wang

2024-05-15

Abstract:For the shape control of deformable free-form surfaces, simulation plays a crucial role in establishing the mapping between the actuation parameters and the deformed shapes. The differentiation of this forward kinematic mapping is usually employed to solve the inverse kinematic problem for determining the actuation parameters that can realize a target shape. However, the free-form surfaces obtained from simulators are always different from the physically deformed shapes due to the errors introduced by hardware and the simplification adopted in physical simulation. To fill the gap, we propose a novel deformation function based sim-to-real learning method that can map the geometric shape of a simulated model into its corresponding shape of the physical model. Unlike the existing sim-to-real learning methods that rely on completely acquired dense markers, our method accommodates sparsely distributed markers and can resiliently use all captured frames -- even for those in the presence of missing markers. To demonstrate its effectiveness, our sim-to-real method has been integrated into a neural network-based computational pipeline designed to tackle the inverse kinematic problem on a pneumatically actuated deformable mannequin.

Robotics

What problem does this paper attempt to address?

The paper aims to address the sim-to-real transfer problem in the shape control of deformable freeform surfaces. Specifically, the core objectives of the study include the following points:

1. **Shape Control Problem**: Investigate how to establish a mapping relationship between driving parameters and deformed shapes through simulation, and use this mapping to solve the inverse kinematics problem, i.e., determine the driving parameters that can achieve the target shape.

2. **Simulation and Reality Gap**: The freeform shapes obtained by existing simulation methods often differ from actual physical deformations due to hardware errors and simplified physical simulations. This paper proposes a new method based on deformation functions to map the geometric shape of the simulation model to the corresponding actual model shape to bridge this gap.

3. **Sparse Marker Data Processing**: Unlike traditional sim-to-real methods that rely on fully collected dense markers, the proposed method can handle sparsely distributed marker points and can still effectively utilize all captured data frames for training even in the case of partial marker loss.

4. **Fast Inverse Kinematics Solving**: By combining neural networks with simulation prediction and sim-to-real transfer, efficient inverse kinematics computation is achieved. Experiments show that this method can quickly generate the required driving parameters on actual soft robot models, thereby achieving precise approximation of the target shape.

In summary, this research is dedicated to developing a new sim-to-real learning method to overcome the challenges of inefficiency and data collection difficulties faced by existing technologies in handling deformable freeform surface shape control, thereby achieving efficient and accurate inverse kinematics solving.