ABSTRACT. Accurate computational models for off-road mobility are crucial in assessing the performance of vehicles operating on deformable terrain. However, the computational cost is very high due to the large dimensional equations required to account for large soil deformation along with the nonlinear granular material behavior. Although an adaptive model order reduction technique using the proper orthogonal decomposition (POD) allows for a significant reduction of the model dimensionality, the projection-based model order reduction necessitates numerous online matrix operations. These calculations hinder the computational benefits of the POD model order reduction. Therefore, this study explores a data-driven modeling approach for eliminating computational bottlenecks of the projection-based POD model order reduction for tire-soil interaction simulations. In the proposed method, the reduced equations are not developed and solved numerically. Instead, Long Short Term Memory (LSTM) neural networks are constructed to predict the time-domain response of the reduced coordinates for the high-fidelity tire-soil interaction model. Such a non-intrusive model order reduction eliminates computationally demanding matrix calculations. Furthermore, the proposed data-driven tire-soil interaction model is integrated into the physics-based multibody vehicle model for off-road mobility simulations. Several numerical examples are presented to demonstrate the simulation capability of the proposed POD-based data-driven tire-soil interaction model. More details on the computational algorithms will be presented at the conference.

ABSTRACT. Accurate control and efficient operation of hydraulically driven systems rely on the precise identification of the friction force within hydraulic actuators. Predicting the friction force is challenging due to its inherent nonlinearities and complex physical nature. This study introduces a data-driven approach using deep neural networks (DNN) to predict nonlinear friction forces. Various modeling techniques for hydraulic actuators exist, with the lumped fluid theory being a widely used method due to its efficiency and accuracy. The LuGre friction model is commonly used to describe friction forces, incorporating variables such as bristle deformation and tangential velocity.The DNN was trained with data from a uniaxial hydraulic actuator and utilized inputs including pressures, actuator length, velocity, and acceleration. The DNN demonstrated its predictive performance in a four-bar mechanism simulation, effectively replacing the mathematical friction model. 

ABSTRACT. In recent years, the System with Rigid body and Extremely Flexible components (hereinafter called “SREF”) is increasingly used in next-generation space crafts which have huge structures [1]. Previous studies have proposed effective methods for analyzing SREF motion, which based on an analogy between the state transitions of the SREF and the contact problem of rigid bodies. Ooshima et al. [2] proposed an efficient analysis method by solving linear complementarity problems for the state transitions of SREF, which include transitions in which the string is subjected to impact tension and transitions in which the tension is lost. In the proposed method, the mass is assumed to be a mass point which has planar motion, and good agreement is confirmed by analysis and experiment. However, since the masses constituting the SREF are not treated as rigid bodies, the effect of their moment of inertia on the motion is not considered. Thus, this paper investigates the effect of moment of inertia on dynamics in tether system using two-dimensional analysis and compare results of analysis and experiments.

ABSTRACT. Combining structural simulations with distributed load models within a co-simulation framework typically necessitates the exchange of substantial sets of input and output data. These data sets, which scale with the size of the corresponding structural finite element mesh, can become a critical factor affecting overall simulation time. To mitigate the data exchange burden in the co-simulation interface, we propose a reduction method based on model order reduction techniques found in small deformation flexible multibody formulations. Consequently, the regarded structural model is subjected to the same limitations as small deformation flexible multibody models. The proposed methodology is implemented using the Flexible Natural Coordinates Formulation (FNCF), resulting in a co-simulation interface reduction strategy that limits the required interface points while maintaining the same level of accuracy.

ABSTRACT. Equipment mounted on warship vessels must be designed to withstand the accelerations caused by underwater explosions (UNDEX) without contact with the hull. Depending on the characteristics of the equipment, its shock endurance is validated on a different shock testing machine. This work studies the dynamic simulation of a shock test on a Medium Weight Shock Machine (MWSM) using multibody methods. The multibody simulation involves flexible bodies that deform under the action of shock (impact) forces. The flexibility is parameterized using the modal reduction technique by means of eigenmodes and static modes under the framework of the Floating Frame of Reference (FFR) method.

ABSTRACT. Mechatronic systems combining rigid or flexible multibody systems with sophisticated control become more and more powerful. Especially AI-methods have been a real game changer enlarging the application field of mechatronics enormously, even to the point where such systems can now act totally autonomously like in autonomously driving vehicles. At the same point, AI-methods may become a game stopper since interpretability of their outcomes is almost lost being in contrast to the demand of high reliability of autonomous systems. What is required is a concept to check that failure probability of such a system is low enough.

The scenario-based validation approach discussed in this presentation meets these challenges by concentrating on safety-critical scenarios. The first step of the concept identifies safety-relevant driving scenarios where it turns out that an AI-based approach called time-series forest (TSF) approach outperforms a state-of-the-art rule-based decision tree. The second step generates artificial driving maneuvers where again the AI-approach based on a variational autoencoder (VAE) outperforms the usual approach of using idealized parametric maneuver descriptions. Finally, modern methods of reliability analysis are combined with AI-based surrogate modeling of the limit state function and SiL-simulation to estimate the failure probability of an level-3 Advanced Driver Assistance System (ADAS) for automobiles.

ABSTRACT. Soft pneumatic actuators (SPAs) are often used in soft robotics due to their low cost and lightweight. However, their compliance and flexibility often lead to nonlinear dynamics, which complicates the design and control of soft robots. This paper deals with the synchronization of multiple SPAs in a soft gripper to ensure the grasping success rate. The synchronization is transformed into an optimization problem with constraints, where the optimal time-dependent input pressures of multiple actuators are determined. The optimized input leads to a synchronized movement of the SPAs and increases the reliability of gripping an object. An example of the synchronization of three different stiff SPAs is used to demonstrate the proposed approach.

ABSTRACT. The worsening effects of climate change in the Arctic call for the need to advance the state of autonomous exploration in the region. As a possible solution, the Multi-terrain Amphibious ARCtic explorer (MAARCO) rover is currently being developed. The MAARCO rover utilizes two helical drives, or Archimedes screws, as its mode of locomotion both on the diverse land terrain of the Arctic and underwater. Since the hydrodynamic effects on the helical drive while travelling underwater are unknown, current dynamic simulations are unable to accurately model the vehicle’s locomotion. This work experimentally determines the propulsive performance characteristics of the helical drive as an underwater propeller using a free-surface water tunnel. The performance characteristics are determined for a set of four different geometries, allowing for a parameter study on the geometric properties of the helical drive. It was determined that the efficiency and thrust generated by the helical drive is more sensitive to the ratio of pitch length to diameter rather than the ratio of total diameter to ballast diameter. Additionally, the determined propulsive performance characteristics suggest that while the thrust produced by a helical drive is similar to that of the modern naval propeller, the torque required is greater, causing the open water efficiency of the helical drive to be lower than a typical propeller. The experimental results are then integrated into a dynamic model of the MAARCO rover using the Newton-Euler method. With the hydrodynamic forces of the helical drive implemented into the equations of motion, the accuracy of the simulations modeling the rover when performing a set of maneuvers underwater is improved.