Fully autonomous earth-moving heavy equipment able to operate without human intervention can be seen as the primary goal of automated earth construction. To achieve this objective requires that the machines have the ability to adapt autonomously to complex and changing environments. Recent developments in automation have focused on the application of different machine learning approaches, of which the use of reinforcement learning algorithms is considered the most promising. The key advantage of reinforcement learning is the ability of the system to learn, adapt and work independently in a dynamic environment. This article investigates an application of reinforcement learning algorithm for heavy mining machinery automation. To this end, the training associated with reinforcement learning is done using the multibody approach. The procedure used combines a multibody approach and proximal policy optimization with a covariance matrix adaptation learning algorithm to simulate an autonomous excavator. The multibody model includes a representation of the hydraulic system, multiple sensors observing the state of the excavator and deformable ground. The task of loading a hopper with soil taken from a chosen point on the ground is simulated. The excavator is trained to load the hopper effectively within a given time while avoiding collisions with the ground and the hopper. The proposed system demonstrates the desired behavior after short training times.

/pdf/automated-excavator-based-on-reinforcement-learning-and-1jvjjc51ap.pdf

Automated Excavator Based on Reinforcement Learning and Multibody System Dynamics


 The effect of friction is widespread around us, and most important projects must consider the friction effect. To better depict the dynamic characteristics of multibody systems with friction, a series of friction models have been proposed by scholars. Due to the complex and changeable working conditions, the contact surface is uncertain, and characterizing the friction properties is a challenging problem. Therefore, in this work, a mechanistic-based data-driven approach is proposed to establish a general friction model. According to the generalization ability of deep neural networks, the proposed strategy can handle the friction in multibody systems with different contact surfaces. Moreover, the proposed mechanistic-based data-driven approach can utilize both numerical data and experimental data, so it can achieve small data to for the dynamic behavior prediction of complex mechanical systems. Eventually, the numerical simulation is compared with the experimental test. The results show that the proposed strategy can predict the dynamic behavior of a complex multibody system well and can reflect many important friction phenomena, such as the Stribeck effect, stiction and viscous friction.

A mechanistic-based data-driven approach for general friction modeling in complex mechanical system

Effective weight-reduction- and crashworthiness-analysis of a vehicle’s battery-pack system via orthogonal experimental design and response surface methodology

Trajectory optimization of an electric vehicle with minimum energy consumption using inverse dynamics model and servo constraints

The undesirable stresses and deformations during a high-speed crash can cause a short circuit or sudden fire in the battery packs, which represents a significant safety concern for vehicles. In thi...

Crush and crash analysis of an automotive battery-pack enclosure for lightweight design

Tree-topology-oriented modeling for the real-time simulation of sedan vehicle dynamics using independent coordinates and the rod-removal technique

Iterative refinement algorithm for efficient velocities and accelerations solutions in closed-loop multibody dynamics

Automotive camera modules are critical components for vision, detection, and recognition in an advanced driver assistance system. They usually consist of housing, a lens, a lens holder, a harness, seals, and printed circuit boards, which are assembled through screws, adhesive, and connectors. Due to a small and complex geometry of the components, the component material failure is a critical issue. In this paper, the lens holder material failure is investigated using the root cause analysis and experimental validation. First, a fishbone diagram model is created to analyze the potential causes and an action checklist is employed to identify the root cause. Second, the structure of the lens holder is optimized according to the finite-element analysis result. The numerical results show that the strength of the optimized lens holder is improved by nearly 35%. Finally, the torque test is carried out to validate the optimized structure experimentally. The presented analysis and validation procedure can be used to solve material failure issues for small-scale automotive components effectively.

/pdf/investigation-on-the-material-failure-in-a-small-scale-433hdn1os8.pdf

Wei Dai

Papers

Tree-topology-oriented modeling for the real-time simulation of sedan vehicle dynamics using independent coordinates and the rod-removal technique

Trajectory optimization of an electric vehicle with minimum energy consumption using inverse dynamics model and servo constraints

Crush and crash analysis of an automotive battery-pack enclosure for lightweight design

Iterative refinement algorithm for efficient velocities and accelerations solutions in closed-loop multibody dynamics

Investigation on the Material Failure in a Small Scale Automotive Camera Module via Root Cause Analysis and Experimental Validation