Search Results

Abstract This paper presents a machine learning application of the force/torque sensor in a human-robot collaborative manufacturing scenario. The purpose is to simplify the programming for physical interactions between the human operators and industrial robots in a hybrid manufacturing cell which combines several robotic applications, such as parts manipulation, assembly, sealing and painting, etc. A multiclass classifier using Light Gradient Boosting Machine (LightGBM) is first introduced in a robotic application for discriminating five different contact states w.r.t. the force/torque data. A systematic approach to train machine-learning based classifiers is presented, thus opens a door for enabling LightGBM with robotic data process. The total task time is reduced largely because force transitions can be detected on-the-fly. Experiments on an ABB force sensor and an industrial robot demonstrate the feasibility of the proposed method.

In tire development, the dynamic tire-road contact forces are an important indicator to assess structure-borne interior cabin noise. This type of noise is the dominant source in the frequency range from 50-450 Hz, especially when rolling with constant angular velocity on a rough road. The spatial force distribution is difficult or sometimes even impossible to simulate or measure in practice. So, the use of an inverse technique is proposed. This technique uses response measurements in combination with a digital twin simulation model to obtain the input forces in an inverse way. The responses and model properties are expressed in the time domain, since it is specifically aimed to trace back the impact locations from road surface texture indents on the tire. In order to do so, the transient responses of the travelling waves as a result of these impacts is used. The framework expresses responses as a convolution product of the unknown loads and impulse response measurements.

The lightweight design of a vehicle can save manufacturing costs and reduce greenhouse gas emissions. For the off-road vehicle and truck, the chassis frame is the most important load-bearing assembly of the separate frame construction vehicle. The frame is one of the most assemblies with great potential to be lightweight optimized. However, most of the vehicle components are mounted on the frame, such as the engine, transmission, suspension, steering system, radiator, and vehicle body. Therefore, boundaries and constraints should be taken into consideration during the optimal process. The finite element (FE) model is widely used to simulate and assess the frame performance. The performance of the frame is determined by the design parameters. As one of the largest components of the vehicle, it has a lot of parameters. To improve the optimum efficiency, sensitivity analysis is used to narrow the range of the variables.

The field performance of a machine is conventionally analyzed using tools of virtual validation such as physics-based simulation models. Machine performance test data is typically not incorporated for performance evaluation using these tools. The present work aims to demonstrate the use of Data Analytics (DA) as a tool to analyze this data for predictive purposes. It aims at establishing numerical relationships of engineering interest within the data, which would otherwise be complex if done only using physics-based models. Engine operation data spanning over three months, comprising of multiple channels, of an off-highway machine, is used for model development. Machine fuel burn rate is chosen as the dependent variable. Several independent variables such as engine speed, charge air pressure, NOx production level, are chosen based on their correlation with the dependent variable and upon engineering interest.

Transport vehicles consume a large amount of fuel with low efficiency, which is significantly affected by drivers' behaviors. An assessment system of eco-driving pattern for buses could identify the deficiencies of driver operation as well as assist transportation enterprises in driver management. This paper proposes an assessment method regarding drivers' economic efficiency, considering driving conditions. To this end, assessment indexes are extracted from driving economy theories and ranked according to their effect on fuel consumption, derived from a database of 135 buses using multiple regression. A layered structure of assessment indexes is developed with application of AHP, and the weight of each index is estimated. The driving pattern score could be calculated with these weights.

The electrification of off-highway machines are increasing significantly. Advanced functionalities, beneficial energy efficiency and effectiveness are only a few advantages of electric propulsion systems. To control these complex systems in varying environments intelligent algorithms at system level are needed. This paper addresses the topic of machine learning algorithms applied to off-highway machines and presents a methodology based on artificial neural networks to identify and recognize recurrent load cycles and work tasks. To gain efficiency and effectiveness benefits the recognized pattern settings are applied to the electric propulsion system to adjust relevant parameters online. A dynamic adaption of the DC-link voltage based on the operating points of the machine processes is identified as such a parameter.

This study investigates the use of machine learning methods for the selection of energy storage devices in military electrified vehicles. Powertrain electrification relies on proper selection of energy storage devices, in terms of chemistry, size, energy density, and power density, etc. Military vehicles largely vary in terms of weight, acceleration requirements, operating road environment, mission, etc. This study aims to assist the energy storage device selection for military vehicles using the data-drive approach. We use Machine Learning models to extract relationships between vehicle characteristics and requirements and the corresponding energy storage devices. After the training, the machine learning models can predict the ideal energy storage devices given the target vehicles design parameters as inputs. The predicted ideal energy storage devices can be treated as the initial design and modifications to that are made based on the validation results.

Autonomous vehicles are currently a subject of great interest and there is heavy research on creating and improving algorithms for detecting objects in their vicinity. A ROS-based deep learning approach has been developed to detect objects using point cloud data. With encoded raw light detection and ranging (LiDAR) and camera data, several basic statistics such as elevation and density are generated. The system leverages a simple and fast convolutional neural network (CNN) solution for object identification and localization classification and generation of a bounding box to detect vehicles, pedestrians and cyclists was developed. The system is implemented on an Nvidia Jetson TX2 embedded computing platform, the classification and location of the objects are determined by the neural network. Coordinates and other properties of the object are published on to various ROS topics which are then serviced by visualization and data handling routines.

As autonomous vehicles continue to grow in popularity, it is imperative for engineers to gain greater understanding of vehicle modeling and controls under different situations. Most research has been conducted on on-road ground vehicles, yet off-road ground vehicles which also serve vital roles in society have not enjoyed the same attention. The dynamics for off-road vehicles are far more complex due to different terrain conditions and 3D motion. Thus, modeling for control applications is difficult. A potential solution may be the incorporation of empirical data for modeling purposes, which is inspired by recent machine learning advances, but requires less computation. This thesis proposal presents results for empirical modeling of an off-road ground vehicle, Polaris XP 900. As a first step, data was collected for 2D planar motion by obtaining several velocity step responses. Multivariable polynomial surface fits were performed for the step responses.

How to improve the measurement accuracy of road gradient is the key content of the research on the speed warning of commercial vehicles in mountainous roads. The large error of the measurement causes a significant effect of the vehicle speed threshold, which causes a risk to the vehicle's safety. Conventional measuring instruments such as accelerometers and gyroscopes generally have noise fluctuation interference or time accumulation error, resulting in large measurement errors. To solve this problem, the Kalman filter method is used to reduce the interference of unwanted signals, thereby improving the accuracy of the slope measurement. However, the Kalman filtering method is limited by the estimation error of various parameters, and the filtering effect is difficult to meet the project research requirements.

Abstract An energy management strategy based on model predictive control (MPC) was proposed for the hybrid bus. For the series configuration, MPC was used for power distribution among transmission components. Real-time optimization of the control strategy was achieved, which improved the fuel economy. First, a rule-based energy management strategy was proposed, and the logical thresholds of the stage of charge (SOC) and the demand power were formulated to underlie the subsequent study of the control strategy. Second, an energy management strategy based on global optimization was established where the dynamic programming algorithm was used to determine the SOC optimal reference curve and the limitation of fuel economy. In this way, the target and reference can be provided for the subsequent control strategy. Third, a radial basis neural network speed prediction model based on wavelet transform was formulated.

Abstract Hybrid Electric Vehicles (HEVs) achieve better fuel economy than conventional vehicles by utilizing two different power sources: an internal combustion engine and an electrical motor. The power distribution between these two components must be controlled using some algorithm, be it rule based, optimization based, or reinforcement learning based. In the design of such control algorithms, it is important to evaluate the impact that variations of certain design parameters will have on the system performance, in this case, fuel economy. Traditional methods of sensitivity analysis have been applied to various power flow control algorithms to determine their robustness to the variations of HEV design parameters. This article presents a sensitivity analysis of three power flow control algorithms: twin delayed deep deterministic policy gradient (TD3), deep deterministic policy gradient (DDPG), and adaptive equivalent consumption minimization strategy (A-ECMS).

Abstract A new method that allows data-enabled (empirical) models, commonly used for automotive engine calibration, to extrapolate beyond the range of training data has been developed. This method used a physics-based system-level one-dimensional model to improve interpolation and allow extrapolation for three data-based algorithms, by modifying the model input (feature) space. Neural network, regression, and k-nearest neighbor predictions of engine emissions and volumetric efficiency were greatly improved by generating 736,281 artificial feature spaces and then performing feature selection to choose feature spaces (feature selection) so that extrapolations in the original feature space were interpolations in the new feature space. A novel feature selection method was developed that used a two-stage search process to uniquely select the best feature spaces for every prediction.

Biodiesels, produced from natural and renewable sources such as vegetable oils are most likely to replace petroleum derived diesel as a CI engine fuel in the long term. However it may be intended to use Biodiesels as blends with diesel in standard proportions. This work makes a thorough analysis of the variation of performance and emission characteristics of CI engine with respect to the proportion of Biodiesel in the blend and also attempts to find the optimal blend depending upon properties of the Biodiesel using Artificial Neural Networks (ANNs).There may exist a particular value of the proportion for every Biodiesel for which the best performance and/or lowest emissions are obtained. Artificial Neural Networks (ANNs) are used for this correlation between percentage of Biodiesel in the blend with performance and emissions. Fuel properties are used as an input to generalize the solution so that the same network can be used for different bio-esters.

Abstract This article describes the application of Reinforcement Learning (RL) with an embedded heuristic algorithm to a multi-objective hybrid vehicle optimization. A multi-objective optimization problem (MOP) is defined as a minimization of total energy consumption and trip time resulting from optimal control of vehicle speed over a known route. First, a computationally efficient heuristic optimization algorithm is formulated to solve the MOP for multiple traffic scenarios. Then, the off-line integration of RL is applied to the heuristic optimization algorithm process and utilized to solve the MOP. Finally, the online optimization capability of the machine learning algorithm is discussed, as well as its extension to the vehicle routing problem and the hybrid electric vehicle. The specific scenario investigated is where a generic vehicle begins a trip on a one-lane highway. The length of the highway and the number of vehicles and traffic signals on the road are generic as well.

Abstract TOC

Abstract Autonomous vehicle technology has become an unprecedented trend in the development of the automobile industry, which can ensure highly efficient use of resources, effectively improve the driving experience, and greatly reduces the driver’s burden. As one of the key technologies of autonomous vehicles, path planning has an important impact on the practical applications of autonomous vehicles. Planning a proper and efficient path is a prerequisite, which can improve the driving experience of autonomous vehicles. Therefore, in-depth research and development on applications of AI technology in path planning definitely have significant value in academic research. In this article, we will introduce a variety of path planning approaches for autonomous vehicles. We summarize the attributes of these path planning algorithms; simultaneously, we analyze the improvements to these algorithms. Then, we have a preliminary discussion on the applications in vehicle positioning and navigation.

Abstract In the past years, the automotive industry has been integrating multiple hardware in the vehicle to enable new features and applications. In particular automotive applications, it is important to monitor the actions and behaviors of drivers and passengers to promote their safety and track abnormal situations such as social disorders or crimes. These applications rely on multiple sensors that generate real-time data to be processed, and thus, they require adequate data acquisition and analysis systems. This article proposes a prototype to enable in-vehicle data acquisition and analysis based on the middleware framework Robot Operating System (ROS). The proposed prototype features two processing devices and enables synchronized audio and video acquisition, storage, and processing. It was assessed through the implementation of a live inference system consisting of a face detection algorithm from the data gathered from the cameras and the microphone.

Abstract TOC

Abstract TOC