Examensarbeten för masterexamen // Master Theses
Länka till denna samling:
Browse
Browsar Examensarbeten för masterexamen // Master Theses efter Program "Biomedical engineering (MPBME), MSc"
Visar 1 - 20 av 21
Sökresultat per sida
Sortera efter
- PostA Parametric Study of Shoulder Belt Interactions with the PIPER Scalable Child Human Body Model in Frontal and Frontal Offset Impacts(2018) Berntsson, Josefine; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Chalmers University of Technology / Department of Mechanics and Maritime SciencesMotor Vehicle Crashes (MVC) are among the leading causes of deaths for children globally. Complementary to the fatality risk, traffic road accidents are responsible for an even larger number of severe pediatric injuries which can lead to permanent impairment. Actions towards reducing or preferably eliminating this problem are to improve the design, accessibility and regulations of child restraint systems as well as implementation of these in actual passenger vehicles. Computer Aided Engineering (CAE) has become an integrated part of the contemporary automotive industry, providing the opportunity to perform crash tests using a simulation platform as a complement to physical crash tests. Finite Element Human Body Models have the potential to become a powerful tool in the virtual assessment of injury tolerances, kinematic behavior and interaction with vehicle interior design. Although substantial efforts have been contributed to the development of male HBM:s, in particular in the 50th percentile size interval, child size models are receiving less attention. Hence, the Position and Personalize Advanced Human Body Models for Injury Prediction (PIPER) project aimed to provide an open source child HBM, scalable within the range of 1.5-6 years old as well as different percentiles within each age group. In addition a scaling and positioning tool, PIPER Framework, has been released to facilitate personizalisation of FE HBM:s. This Master's Thesis project has evaluated the sensitivity of the PIPER scalable child model by exposing the PIPER baseline model (6 year old, 50th percentile) to three different shoulder belt angles, achieved by altering the D-ring position in an LS-DYNA environment. These belt routing setups have been applied to three different cases which included a belt-positioning booster, a high-back booster and a rear seat only during frontal and frontal offset impact. Kinematic behaviour of the child HBM and the occurrence of undesirable belt performance such as sliding of the shoulder resulting in increased roll-out risk or neck-loading, has been the main evaluating feature. Head accelerations, de ection of sternum and global head injury criteria (HIC) have been estimated and additionally used for comparison between the different parameters. Conclusions were made that the PIPER scalable child model was able to capture different undesired belt interaction behaviours such as roll-out and submarining, although further evaluations are necessary. A visual comparison with data retrieved during the literature review supported the exible kinematic trajectories of the PIPER scalable child HBM when exposed to the high-severity crash pulses of the parametric study. In terms of injury criteria, the PIPER scalable child HBM appeared to overestimate HIC15 values, hence additional research concerning biofidelic injury thresholds for children is necessary.
- PostA study of sitting posture and belt position in a travelling car: How do passengers sit in a travelling car?(2019) Hansson, Annika; Lysén, Emma Nilsson; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Chalmers University of Technology / Department of Mechanics and Maritime SciencesTo improve the future design of restraint systems, it is important to know how passengers’ sitting postures change over time and how the passengers interact with the restraint systems. This master thesis at Chalmers University of Technology, focuses on pelvis rotation, slouching and belt position while travelling in a front seat of a car on regular roads. The information was collected during normal drive in the passenger seat of a regular car, while the volunteers perform activities such as; resting, e-socializing and conversing. Twenty volunteers, ten male and ten female, participated in the study. Volunteers were seated in the front row passenger seat, because this sitting posture is probably similar to how passengers will sit in future autonomous cars. The inertial motion measurement system MTw Awinda from Xsens, in total eight sensors, were used in the study. They were placed on the volunteer’s sacrum, sternum, C7, T3, L5, forehead and car. The data from the sacrum sensor, that corresponds with pelvis rotation are mainly presented and discussed in this report. In addition, a surface pressure sensing array (Tekscan mat) was placed on the car seat cushion. The data from selected volunteers was analyzed with the TEMA to determine degree of slouching. Photo analysis was carried out to assess belt positions before and after the test. Additionally, the rotation of pelvis and sternum when changing seat back angle in intervals of 5° between 23° and 48° were also investigated. The results show that pelvis rearward rotation increases by average 10° when riding in the car for about 45 minutes. Comparing the activities, the volunteers had similar average pelvis rotation. Slouching could be measured only for three volunteers out of 20 and it seemed to increase on average 3 cm during the ride. The belt position of initial and final sitting posture indicates that the diagonal belt moved less than the lap belt. To investigate the dynamic belt position, future video analysis is needed. Increasing seat back angle appeared to have a correlation with increasing sacrum and sternum pitch.
- PostAutomatic detection of saccadic eye movements using EOG for analysing effects of cognitive distraction during driving(2017) Cafasso, Alessandro; Karlsson, Sebastian; Chalmers tekniska högskola / Institutionen för tillämpad mekanik; Chalmers University of Technology / Department of Applied MechanicsDriver distraction is a relevant driving safety issue and an ongoing field of research. A particular distraction is cognitive distraction, which refers to when the driver is mentally engaged in a task unrelated to driving, e.g. talking to a passenger. Eye movements can be analyzed to study effects of cognitive distraction during driving, and are typically recorded using video-based eye tracker systems. An alternative technique that might be suitable for eye movements measurements during driving is the electro-oculography (EOG). EOG is a method for recording the electrical signal of the eyes as they move. One interesting eye movement in cognitive distraction studies is the saccade, the rapid movement of the eye from one point of interest to another. The primary purpose of this thesis is to develop an algorithm for automatic detection of saccades using EOG. The resulting algorithm is a combination of two modified existing eye detection algorithms, namely Continuous Wavelet Transform Saccade Detection (CWT-SD) and Shape Features. It is found that the developed algorithm can be used in driving environments if good signal quality can be assured. The secondary purpose of this thesis is to investigate how cognitive distraction affects saccadic rate and amplitude during driving. The findings suggest a statistically significant decrease in saccadic rate during cognitive load but not in saccade amplitude. However, further research on bigger datasets and different driving scenarios is needed to verify the results.
- PostBiofidelity Evaluation of Thoracolumbar Spine Model in THUMS(2016) Afewerki, Henok; Chalmers tekniska högskola / Institutionen för tillämpad mekanik; Chalmers University of Technology / Department of Applied MechanicsThoracolumbar spine injuries in motor vehicle crashes are occurring and the mechanisms are poorly understood. It has been hypothesized to be connected to vehicle’s restraint systems but further studies are required to understand and subsequently address the problem in future restraint systems. Finite Element (FE)-Human body models are invaluable tools for crash analysis, however, quality of the response depends on the biofidelity of the model. The objective of this thesis is to evaluate biofidelity of the thoracolumbar spine model in Total Human Model for Safety (THUMS), Toyota Motor Corporation and Toyota Central R&D Labs . In this thesis work three dynamic and one static thoracolumbar experiments were simulated. THUMS’ ligaments were verified against cadaveric data. Two modified disc material models were inserted in to THUMS and the results compared against experimental data. The Global Human Body Model Concertium model (GHBMC), GHMBC, LLC was also evaluated against cadaveric data from two experiments. All simulations were run in LS-DYNA and pre and postprocessing tasks were performed in LS-PrePost and Matlab. The response of the lumbar FSUs in THUMS’ under the dynamic compression test was similar to the experimental data but was three to four times less stiff. On the other hand, the T12-L5 segment showed fair correlation of reaction force whereas reaction moment was significantly lower. Kinematics of the cadaveric spine under flexion and extension tests was not captured. Reaction moment, shear force and vertical displacement were found to deviate from the response of the cadaveric specimens during the dynamic flexion and shear test. Only horizontal displacement showed good correlation in this test. THUMS performance was good in the static flexion and shear test but poor in flexion only test. Furthermore, the Capsular Ligaments (CL) and the Ligamentum Flavum (LF) in THUMS were found to be about three times shorter and stiffer, respectively. In all the simulations the intervertebral contacts were responsible for the sudden and large increase and vibrations occurring at about the experimental failure point. The modified disc material models improved response of only the lumbar FSUs under the compression test. In conclusion, biofidelity of the thoracolumbar spine model in THUMS is found to be poor and remodelling is necessary. The compliant nature of the intervertebral discs, the shorter length of the CL and higher stiffness of the LF and the smaller initial invetervertebral gap were identified as the main weaknesses of the model.
- PostCar occupant seat belt fit; the effect of belt pre-pretensioning(2023) Bohl, Louise; Eliasson, Klara; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Chalmers University of Technology / Department of Mechanics and Maritime Sciences; Davidsson, Johan; Hederskog, Amanda; Östling, MartinIn the event of a crash, the seat belt should load the occupant’s pelvis, thorax, and clavicle. A shoulder belt segment routed distal of the shoulder, i.e. positioned on the arm, may cause chest and abdominal injuries during a crash. The overriding aim of this study was to investigate if an improperly positioned shoulder belt can be repositioned to a proper position on the clavicle, with the help of a pre-pretensioner for front seat occupants. More specifically, the aims were to investigate if the location of the belt attachment points, occupant body characteristics, belt geometry, belt fit, and friction of clothing affected the ability of the pre-pretensioner to reposition the shoulder belt and from which distances down the arm it was possible. A volunteer study was conducted to investigate if the shoulder belt could be repositioned for a nominal belt geometry similar to a Volvo S60 and for a belt-in-seat geometry in an adopted test rig. Several anthropometric, belt geometry, and belt fit measurements were collected and analyzed to identify why the belt did not reposition for some individuals compared to others. In addition, the ability of the prepretensioner to reposition the shoulder belt for different fore-aft seat positions, Dring heights, and a belt-in-seat installation were investigated. 17 male and 18 female volunteers were tested. The study found that the location of the belt attachment points affected belt repositioning, since the shoulder belt was not repositioned for the majority of the volunteers in the belt-in-seat installation. The belt repositioned for all volunteers in the most common seat positions while the rate of unsuccessful repositionings increased for more forward seat positions. A high D-ring made belt repositioning possible for all volunteers. Measurements identified as influencing belt repositioning were a taller shoulder height (measured while seated) and a smaller abdominal depth in seat positions forward of the mid position and for the belt-in-seat installation. The belt did not reposition with the lower friction clothing material in the belt-in-seat installation but repositioned for some in a forward fore-aft position. The repositioning commonly failed from positions close to the acromion on the arm. The results indicate that the upper body shape influences belt repositioning. It could be linked to combinations of upper body measurements, shoulder belt routing, and different belt geometries. Based on the results, future studies should investigate shoulder belt repositioning for additional belt geometries and dynamic scenarios as well as the possibility to implement belt fit warning interventions.
- PostClustering and geometrical features for classi cation of humans in three dimensional data(2019) Hedin, David; Wendel, Johan; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Boyraz Baykas, Pinar; Pålsson, AlbinComputer vision is a rich research field that uses images from normal or technically specific cameras to perform tasks ranging from surveillance to autonomous driving. Computer vision with depth images are however relatively new. Depth images add a third dimension to the image by giving every pixel a depth value and can be produced with several different camera types. The implications for depth imagery is that if an object can be classified in a depth image as e.g a trashcan, a cat or a human, the nature of the data immediately also gives us the distance and position of that object. The extra data from these cameras can enable estimating the volumes and sizes of objects in an image with some extra processing. The focus of this master thesis is on the processing and analysis of this depth data to enable object identification and human classification. Much research has been on analysing 3D data from the 2D perspective, in this thesis the captured data is first converted to Cartesian coordinates before attempting classification, yielding further possibilities. The goal of this thesis was to find if there are some anthropomorphic-geometrical features that can describe the human body well enough to accurately classify humans in the Cartesian data. The features are used in two ways, as a Heuristical-geometrical filter and as features for a support vector machine. Furthermore the thesis presents a successful dynamic adaption of the fast-DBSCAN (Density Based Spatial Clustering of Applications with Noise clustering) algorithm for 3D Cartesian data and a slice method for finding local maxima of point cloud objects. The results show that anthropomorphic-geometrical features can to an extent be used to classify Cartesian point cloud data. Low resolution cameras has potential for classification purposes as resolution seem to have little effect on geometrical classification as long as human resolution is no less then 20px vertically. Some further work would be needed to create a anthropomorphic-geometrical for real world application.
- PostDeveloping a Protocol for Detection of Axonal Injury in Rats Using Diffusion Tensor Magnetic Resonance Imaging - Evaluating large volumes of data to find microscopic injuries(2015) Kvistberg, Hanna; Lundgren, Martin; Chalmers tekniska högskola / Institutionen för tillämpad mekanik; Chalmers University of Technology / Department of Applied MechanicsTraumatic brain injury is a serious injury that is all too common. Commonly traumatic brain injury includes stretching of the axons which is due to the movement of the brain. Today it is not possible to detect axonal injury using conventional imaging techniques, but the possibility to do so would provide superior screenings for injury and guide treatment of patients. The goal of this thesis was to examine the possibility of creating a new protocol for detection of such injuries. Diffusion tensor magnetic resonance images from animal studies on rats, five exposed and four normals, have been studied. The main focus of this thesis was to attempt to differentiate the datasets and find injuries located in the corpus callosum. The study began with examining the results of a previous study carried out on the same dataset. This examination was followed by evaluating fractional anisotropy and tractography, with the conclusion that fractional anisotropy can be useful but possibly insufficient for detection of axonal injuries when the spatial resolution is too high. Since fractional anisotropy stays the same in points where all eigenvalues increase or decreases equally, injuries might go undetected. Tractography suffers from a few problems underlying with DT-MRI itself, namely the poor ability to resolve crossing fibers, but can be a useful tool in smaller brain regions. Further attempts at spatial normalization and amplitude normalization was made to facilitate a voxel-wise analysis of the brain, comparing each voxel of the traumatized and aligned brains with the voxels of healthy animals. The results did not show significant differences between the healthy and injured animals. Attempts at using classifiers to differentiate between the animals were made, using both neural networks and a linear discriminant analysis classifier. For the neural network classifier this seems promising, but the number of animals in the study was not large enough to be able to perform a complete evaluation. Suggestions on how to improve detection of axonal injuries and verifying the possibility to do so are presented; such as using phantoms to be able to place controlled injuries simulating axonal injuries and comparing images before and after introducing the injury. Such an approach would facilitate detection of diffuse axonal injury by ensuring that a detected difference stems from the injury alone.
- PostDefining and Evaluating New Load Cases in Autonomous Cars(2016) Jorlöv, Sofia; Chalmers tekniska högskola / Institutionen för tillämpad mekanik; Chalmers University of Technology / Department of Applied MechanicsAutonomous vehicles (AVs) are the focus for many research projects right now and are widely regarded as the most important next step for the automotive industry. AVs are anticipated to result in safer roads and with fully automated vehicles the driver does not need to be involved in the driving. This new situation opens up the possibility for new spatial orientations within the vehicle, which might be a challenge for today’s passive safety systems. This master thesis is done in collaboration with Autoliv and aims to define new potential load cases in fully automated cars and to evaluate how today’s passive safety systems handle these new load cases. A literature review were complemented with a qualitative study to understand what activities occupants expect they will be engaged in when no one needs to focus on the driving, and how they will be seated. The qualitative study was performed at a local exhibition called V˚arg˚ardam¨assan with a method called ”Setting the Stage”, developed specifically to explore user expectations. To evaluate the passive safety system, twelve sled tests with the THOR dummy were run in three of the newly defined load cases. A position representing the extended living room, with the front seats rearward-facing, and a relaxing position, with the seatback reclined, were the most popular new seating positions both in the literature and in the qualitative study. These positions together with a conversation position, with the fronts seats rotated 30°inboard, were run in sled tests. The test matrix included rear-end impact, frontal impact and three different speeds (24 kph Euro NCAP rear-end pulse, 40 kph and 56 kph FMVSS 208 full frontal barrier pulse). The restraint parameters differed for each test and consisted of a driver airbag, an extra belt, an simulated knee airbag, a PRC (Pelvis Restraint Cushion), load limiting, pretensioning and modifications of belt geometry and seat adjustments. The rearward-facing living room position generally performed well in frontal impact (rear-end impact for the dummy), with low injury risk values for HIC, BrIC and femur. The challenge for this position is to absorb the energy in the seat construction. The relaxing position was more challenging and got more focus in frontal impact; a beneficial kinematic for keeping the dummy in position, without high spinal loads, was an immediate forward motion or rotation of the upper body. The extra belt was the most important restraint system for the conversation position in frontal impact, but new solutions might be found that could handle this position. Only a few tests were performed for this load case, but much can be learned from small overlap/NHTSA oblique testing that resemble the test carried out in the conversation position. The method ”Setting the Stage”, used in the qualitative study, resulted in detailed and realistic reflections from potential occupants and five new load cases were defined in total in the first part of the thesis. Some load cases were evaluated using sled tests, which limited the test matrix due to simplifications in the setup and limitations of the ATD; still, useful information was gained. Beneficial kinematics were identified and future potential challenges addressed for the two most popular load cases: the relaxing position and the living room position. This thesis was a first exploratory attempt to understand potential challenges within passive safety in autonomous vehicles, but more research and testing is needed to complete the results.
- PostEye Blinks as an Indicator of Car Drivers’ Visual Attention: A statistical analysis of differences in eye blinks between roads of high and low complexity(2021) Ek, Paula; Österberg, Felicia; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Bärgman, Jonas; Nilsson, EmmaThe majority of all traffic crashes occur due to human error, according to National Highway Traffic Safety Administration. As the development of self-driving vehicles´ progress, the driver's role changes to a more monitoring nature, successively eliminating the effect of human error. However, until fully automated vehicles are achieved, the driver needs to be ready to take control over the vehicle in critical situations. Therefore, visual attention is an important attribute to be a reliable car driver engaged in the traffic environment. The thesis sets out to investigate the usability of human eye blinks as an indicator of car drivers´ visual attention. The investigation is based on electrooculography (EOG) measurements obtained during an on-road experiment performed by Volvo Cars and Research Institutes of Sweden (RISE). The data is used to analyse differences in blink rate and half blink duration between interchange (high demand of attention) and motorway (low demand of attention). Surprisingly, the results indicate that the blink rate increases during interchanges, which contradicts findings from previous studies. The contradiction derives from an increase in blink-saccadic pairs occurring due to the driving behaviour in interchanges. \\ Additionally, the result implies that half blink duration increases during motorways. From the findings, it is concluded that blinks without large saccadic eye movements are less affected by driving behaviour and could therefore be a potential robust indicator of visual attention. Further, half blink duration is a possible indicator to measure a drivers' visual attention. Therefore, eye blink measurements have the potential to alert vehicle safety systems about the driver´s level of engagement.
- PostFluid-Structure Interaction Analysis of Carotid Artery Blood Flow. A Patient-Specific Investigation of the Impact of Arterial Wall Deformation on Hemodynamics.(2022) Nordenström, Simon; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Ström, Henrik; Brüning, JanCardiovascular disease, predominantly caused by atherosclerosis, is the leading cause of death worldwide. Therefore, scientific methods capable of accurately evaluating risk factors of cardiovascular disease on a patient-specific basis are highly soughtafter. Typically, numerical approaches for simulating blood flow assume arterial walls to be rigid structures. However, to obtain more physiologically realistic simulations, the interplay between wall deformation and blood flow, the so-called fluidstructure interaction (FSI), must be taken into consideration. In this project, a workflow is established for performing patient-specific FSI simulations of blood flow and arterial wall deformation in the carotid artery – the main source of blood supply to the brain. The workflow is primarily intended to aid in the execution of FSI analyses in the STAR-CCM+ simulation software. Furthermore, the impact of FSI on hemodynamic parameters such as flow velocity, wall shear stress, and oscillatory shear index, is quantified and analyzed for a patient-specific geometry. It is shown that a rigid wall assumption results in underestimations of the areas associated with an increased risk of atherosclerosis, namely regions of low time-averaged wall shear stress and high oscillatory shear index. Moreover, the rigid wall assumption leads to overestimations of the flow velocities during the systolic phase of the cardiac cycle. FSI analyses are performed using three different material models: an isotropic linear elastic model, a Neo-Hookean model, and a Mooney-Rivlin model. For the specific boundary conditions and parameter values employed, marginal differences are observed in the hemodynamics between the three material models.
- PostImprovement of Human Body Model Rib Fracture Risk Prediction: Creation of injury risk curves and rib cortical bone regression models for age adjusted risk prediction(2020) Blennow, Amanda; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Davidsson, Johan; Larsson, Karl-Johan; Lübbe, NilsIn motor vehicle collision, one of the most commonly injured body regions is the thorax and these injuries are in many cases the cause of death among belted, adult drivers. Rib fractures are particularly dangerous among elderly and to enable injury prevention, proper risk assessment is of great importance. In a recent study, the effect of age, sex and strain rate on the rib cortical bone material was investigated, enabling creation of new injury risk curves and rib cortical bone material models.The aim of this thesis was thus to improve the human body model rib fracture risk prediction by creating new rib fracture risk curves and material models of human rib cortical bone for individually adapted rib fracture risk prediction. Using the methods of parametric survival modelling, injury risk curves for rib fracture risk prediction were developed for prediction using both ultimate strain and total strain energy density. It was found that the new strain based risk curve corresponds to a decrease in ultimate strain of 12 % per decade of life, compared to the 5 % per decade of life in the currently used risk curve. Regression models of the mechanical properties elastic modulus, plastic hardening modulus and yield stress, of human rib cortical bone were created and thus, construction of age dependent stress-strain curves was enabled. The stress-strain curves also model the individual variability, suggesting mechanical parameter values for weak, average and strong individuals of any given age. The material models were implemented in the SAFER HBM and four post mortem human subject sled tests were reproduced and outputs were evaluated with both the newly created injury risk curves and the injury risk curves that have been used up until now. It was found that both the new injury risk curves and the material models affected the predicted injury risks and particularly among elderly. Due to the large set of data used for the development of injury risk curves, the new strain-based risk curve possesses a higher statistical validity, compared to the old strain-based risk curve, and can therefore be considered as a better alternative. The energy-based risk prediction proved to be less sensitive to the material model used compared to the strain-based risk prediction . This is promising in terms of predicting risks that are less sensitive to individual variability.
- PostKinematics and shoulder belt position of child volunteers when exposed to steering manoeuvres in different restraint systems(2013) De Faveri, Elisa; Chalmers tekniska högskola / Institutionen för tillämpad mekanik; Chalmers University of Technology / Department of Applied MechanicsIn car crash scenarios, the head is the most frequently injured body region among children. The main injury mechanism for rear seated restrained children, aged 3-13, who sustained head injuries rated AIS2+, has been found to be the contact of the head with the seat back or with other parts of car interior. Previous studies showed that pre-crash manoeuvres could influence the injury outcome. This thesis quantifies the static belt measurement and the kinematics of child volunteers exposed to steering manoeuvres when restrained on different types of booster cushion. A study was conducted on a test track with 18 children divided into two groups based on their stature. Each child was tested for the static belt measurement on three different booster cushions and performed two steering manoeuvres on each of two of these boosters. Cameras were used to monitor the child during the event. Different parameters were extracted from the videos in order to analyse the kinematics of the child. From the static belt measurement arose that tall children can achieve a better belt fit than short children, especially in terms of shoulder belt position and its “grabbing” effect.The analysis of the kinematics showed that head and torso moved inboard in different ways. The position of the belt on the shoulder was also considered throughout the steering event. The belt slip off the shoulder in 11 trials out of 18 for short children when restrained on the accessory booster, while the slip off occurred only in 2 trials when restrained on integrated booster. For tall children no belt slip off occurred regardless of the type of booster cushion.
- PostModeling of Occupant Kinematic Response in Pre-crash Maneuvers A simplified human 3D-model for simulation of occupant kinematics in maneuvers - A simplified human 3D-model for simulation of occupant kinematics in maneuvers(2018) Cyrén, Oscar; Johansson, Sofia; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Chalmers University of Technology / Department of Mechanics and Maritime SciencesMost car manufacturers today equip their cars with collision avoidance systems which can act to avoid or mitigate a crash. Retrospective studies have shown that these systems help reduce the number of crashes, however the intervening evasive maneuvers provoke occupant displacements. Consequently the occupant’s position relative to the restraint systems of the car is affected. Therefore, there is a need to investigate occupant safety also for the pre-crash phase when these evasive maneuvers take place. There are an extensive number of maneuvers to consider and the existing methods for simulating the occupant during the pre-crash phase are inefficient with respect to time. For that reason, a more efficient dynamic model of the occupant was developed to primarily compute the head kinematics and secondarily T1 kinematics, when the occupant model is subjected to the linear- and rotational accelerations that are induced by evasive maneuvers. The model represents the upper body of an occupant, i.e. a front seat passenger or a driver, which is restrained by the seat, the seat belt, as well as the arms if the model is a driver. Two types of occupant models were proposed based on the inverted spherical pendulum theory. The systems of differential equations were derived using Lagrangian mechanics and implemented in Simulink. The models were tuned and validated based on pre-existing volunteer data from vehicle maneuver studies. The results demonstrate that the models were able to capture the occupant kinematics by showing similar dynamic behavior to the kinematics of test subjects in volunteer tests. The computation time when simulating a maneuver of approximately 4 seconds, resulted in a computation time of 0.7 seconds for both models, which allows for efficient computation of an extensive number of pre-crash maneuver simulations for analysis of occupant kinematics.
- PostPredicting Finite Element Simulation output using Machine Learning: A study to understand the potential of Graph Neural Networks in predicting vehicle occupant pre-crash simulation kinematics(2022) Fichera, Chiara Rosanna; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Chalmers University of Technology / Department of Mechanics and Maritime Sciences; Iraeus, Johan; John, JobinEach year around 1.35 million people die in traffic crashes and many more get injured. Human Body Models have become a popular tool to understand and prevent injuries and fatalities as they simulates the human body response during crash sce narios. With improvements in the accuracy and bio-fidelity of Human Body Models, the demand for computational resources is increased up to requiring days for a pre single crash simulation. Therefore, Dimensionality Reduction methods can be used to create a surrogate that is a smaller representation of the model. The reduced representation translated to a latent space that still retains the complexity and the features of the simulation. To build the surrogate, supervised learning tasks have been implemented on the reduced dimension to map the simulation to its parameters. To create a compressed version of the input simulation, Graph Neural Networks have been considered since they incorporate the geometrical structure of the model. The Graph Neural Network has been compared to an equivalent Convolutional Neu ral Network architecture and to a Principal Component Analysis. Random Forest, Gradient Boosting, and XGBoost were chosen and compared to build the surrogate model. Due to computational limitations, and since pre-crash simulations are al ready time-consuming, a simplified Hybrid III dummy model was chosen in place of a full Human Body Model. For all Dimensionality Reduction methods, the accuracy of the reconstruction im proves by increasing both the number of samples and the latent space size. Principal Component Analysis shows a better performance in terms of lower errors compared to Graph Neural Network. Moreover, the Graph Neural Network structure is compa rable to an equivalent Convolutional Neural Network architecture in term of perfor mance. Random Forest and Gradient boosting proved to be better than XGBoost. Principal Component Analysis requires less computational time and resources. Although Graph Neural Network was outperformed in this study, further improve ments in the development of the method may still have potential.
- PostPupillary Measures as Indicators of Cognitively Versus Automatically Controlled Processes(2018) Granberg, Sofia; Wallhede, Malin; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Chalmers University of Technology / Department of Mechanics and Maritime SciencesThe Cognitive Control Hypothesis is a hypothesis that states how car drivers are affected by cognitive load. It says that ”Cognitive load selectively impairs driving subtasks that rely on cognitive control but leaves automatic performance unaffected”. To test if this hypothesis is true, a method for measuring both cognitively demanding and automatized tasks is needed, but also a method for distinguishing between these two conditions. This thesis was conducted at the Vehicle Safety Division at Chalmers Univeristy of Technology together with Volvo Cars, with the aim to investigate if pupillary measures can act as indicators of cognitive load. The aim was also to investigate what eye measures are the most reliable and/or preferable for assessing cognitive load. The research questions were what eye/pupillary measures that can be used in order to measure cognitive activity and if these pupillary measures can be indicators of whether a task is cognitively demanding or automatically performed. The aim was also to develop an experimental method where the data of pupillary measures was to be collected with an eye tracking camera. The project did not investigate other physiological measures and no naturalistic driving was performed. The main experiment included a method where participants performed a motor piano task consisting of different key pressing combinations, performed at two sessions. The task also had a sequence where math questions were induced in order to vary the mental workload. The motor task was expected to (to some extent) be automatized at the second session. The most important finding was a 10Hz peak in the frequency domain, mostly visible during the sequences where the participants were asked math questions. The results also showed that the energy content within 3.75-15Hz was higher for sequences with math questions and thus indicating a higher mental workload. It was also observed that the pupil behavior varied a lot, both within and among the individuals. No significant pattern was observed regarding pupil dilations within and among individuals when they were exposed to math questions.
- PostQuantifying Drivers' Behaviours when Overtaking Bicyclists on Rural Roads - A Study Using Naturalistic Driving Data from a Vehicle's Perspective(2017) Nero, Gustav; Chalmers tekniska högskola / Institutionen för tillämpad mekanik; Chalmers University of Technology / Department of Applied MechanicsOvertaking manoeuvres of vulnerable road users on rural roads have previously been found to be accident-prone events with severe or fatal outcomes for the involved vulnerable road users. Most of the previously conducted studies revolving around overtaking manoeuvres of vulnerable road users on rural roads have used either naturalistic driving data which was collected from the bicyclists' perspective or data from driving-simulators. The work described in this thesis involved quantifying data of cars overtaking vulnerable road users on rural roads. The thesis exemplified how to extract overtaking manoeuvre segments from naturalistic driving data and further demonstrated how to extract comfort zone boundaries. The data used was extracted from the database of the European Naturalistic Driving project UDRIVE and included CAN-data, Mobileye-data, and video data. The video data came from cameras capturing both the inside and the outside of the ego vehicle. The data was enriched via manual annotations and automatic derivation of signals using tools such as SALSA and MATLAB. Manual annotations also verified the data, since not all data extracted from the database contained overtaking manoeuvres. To keep the work manageable only events where a single vulnerable road user traveling in the same direction as the ego vehicle and where the vulnerable road user was in the outer-most lane as the ego vehicle were considered. The focus of the thesis has been method development, that is by primarily using ME data identifying overtaking segments from the UDRIVE database and then derive comfort zone measures such as time to collision, lateral clearance and minimum distance. However, due to issues with A) Subjectivity of the video annotations, B) the lack of a comprehensive quality check of the data (i.e. not comparing what the video-feed showed with what various signals implied), and C) error in a derived measure (i.e. the speed of the vulnerable road user), which in turn was used in several other derived measures, results were only compared to previous studies briefly. For future work the quality of the data (both raw and derived) should be considered to have a higher priority. In other words, a more comprehensive data validation should be performed to verify that extracted data is fit for analysis.
- PostQuantitative analysis of rear-end crash causation mechanisms based on naturalistic crash data(2016) Eiríksdóttir, Hrafnhildur Hekla; Chalmers tekniska högskola / Institutionen för tillämpad mekanik; Chalmers University of Technology / Department of Applied MechanicsUntil recently, little has been known about what exactly happens in the seconds leading up to a car crash. Due to the emergence of naturalistic driving data, e.g. video data of the forward roadway and of the driver combined with various sensor readings from real traffic incidents, it is now possible to research underlying crash causation mechanisms with much greater detail. An analysis of a 100 different rear‐end events, 70 crashes and 30 near‐crashes, was performed with the aim of replicating the findings of the SHRP2 naturalistic driving study performed by SAFER. The findings were in correspondence with those of the SHRP2 study; that rear‐end crashes usually occur due to a combination of glance duration and change rate of the situation kinematics, and that a short glance usually requires a rapid change in the situation kinematics while a longer glance could cause a crash even if the kinematic situation changes relatively slowly. The key mechanism behind crashes was found to be the timing of the last glance off the road relative to the change in urgency, represented optically by looming cues, during the glance. Brake lights were frequently ignored and the act of missing the brake light onset (BLO) in itself was not found to be a key mechanism in causing crashes and near‐crashes. Drivers that ended up in a crash were twice as likely to have looked away from the road after having seen the last BLO as those who ended up in a near‐crash.
- PostThe influence of isometrically derived neck muscle spatial tuning patterns on head response in dynamic conditions(2017) Ivančić, Andrea; Pradhan, Vikram; Chalmers tekniska högskola / Institutionen för tillämpad mekanik; Chalmers University of Technology / Department of Applied MechanicsThe Chalmers AHBM was setup in LS-DYNA to mimic maximum voluntary contraction (MVC) experiments, and to find the optimum cervical muscle activation parameters for a setup using an optimization model in LS-OPT. The setup in LSDYNA incorporated a discrete beam element with a high stiffness, attached to the head of the AHBM, making it resistant to change in length. Pulling against this beam is similar to pushing against a rigid clamp used in the MVC experimental setups with human volunteers. The AHBM head is constrained in 3D space to have small translational and rotational displacements. Hence, it was considered an apt analogy in the numerical simulations of the AHBM, towards obtaining MVC. A linear response surface was used to obtain optimum values for muscle activation parameters in the design space, with sequential domain reduction and tolerance of convergence set to ±1%. The motion of the models was post-processed in LSPrePost. Muscle activation patterns obtained from optimizations for the two models - one with 9 muscle groups and the other with 13 muscle groups in the neck, were compared with the MVC experiment conducted by Siegmund et al. (2007), and later used for simulation of the AHBM with applied dynamic load in 5 different directions, 0, 45, 90, 135 and 180, using a previously implemented 1 DOF PID controller for the AHBM by Östh (2014) as well as a newly developed 3 DOF PID controller, which was based on functions in LS-DYNA. The dynamic spatial tuning patterns for the optimization models were compared to those obtained from the human volunteer sled experiments conducted by Ólafsdóttir et al. (2015). The muscle activation patterns from the optimizations showed great variation compared to those obtained from the MVC experiments, due to difference in muscle modelling, constraints on the AHBM and the optimization setup. The dynamic spatial tuning patters for the optimization setups with 1 DOF PID controller were scaled in the same directions as the muscles were activated in the MVC optimizations, however, the 3 DOF PID controller behaved differently. Larger translational and rotational head displacements were observed in the optimization models as compared to the experimental model. Models with 3 DOF PID controller resulted in significantly different kinematics as compared to the models with 1 DOF PID controller.
- PostTowards Improvement of Human-Machine Interaction: Design of Multimodal Human Intent Recognition System(2019) Danauskiene, Asta; Machado, Mauricio; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Boyraz Baykas, Pinar; Boyraz Baykas, PinarThis master thesis focuses on investigating the electrical brain activity, eye gaze and pupil behaviour in the scope of goal-directed movement intention recognition for human-machine interaction applications. Previous studies support that the electroencephalography (EEG) data is suitable for early motion recognition and prediction and the pupil size changes correlate with the difficulty of the task. However few studies have looked into neural correlates of goal-directed and no-goal movements as well as the correlation between the pupil changes, EEG data and hand motion. We explore these questions through a set of cue-based movement experiments that include changing goal, repeating goal and no-goal scenarios and are performed in collaboration with a robot. The results were analysed with regard to movement related cortical potentials (MRCP) and event related spectral perturbation (ERSP) of EEG data, evoked pupil response, gaze patterns as well as binary goal\no-goal classification of the data and correlation between different biosignals. Our results indicate that changing goal-directed movements are distinguishable from no-goal movements in EEG data in both temporal and time-frequency domains, when performing the task with a passive robot. Collaborative robot experiments showed great intersubject variability, therefore need to be further investigated. No correlation between evoked pupil response and MRCP was found in this study, however results suggest a correlation between MRCP and motion velocity profile.
- PostWave propagation in human tissue from transient high frequency vibrations(2020) Mathiyalagan, Priyanka; Nadafan, Parsia; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Berbyuk, Viktor; Johansson, Håkan; Lindell, HansHand-arm vibration injury from the daily exposure of vibrating hand-held machine is one of the common health injuries and causes severe nerve cell damage to the person operating the machine for a long time. The major health problems caused by daily use of vibrating tools causes signs and symptoms like peripheral vascular and peripheral neural disorders of the fingers and hands. These disorder symptoms include numbness, pain, and blanching of the fingers. ISO 5349 claims that the vibrating machines with frequencies in the range of 8-1250 Hz must be regulated. However, vibrations with frequencies higher than 1250 Hz from machines like an impact wrench or dental drills can also damage nerve cells and blood vessels. The objective of this thesis is to study wave propagation in human tissue from transient high frequency vibration of machines tools. The software LS-DYNA is used for performing the simulation and analyze the results. The 2D cross-section finite element model is created for the simulation to see the wave propagation into the finger tissue. The following analysis of simulation results shows that the different load cases from the hand-held vibrating machines affect the skin. However, the change in bulk modulus and decay constant does not make much difference. The intermediate skin layer has more impact caused by the hand-held vibrating machines.