Mekanik och maritima vetenskaper (M2) // Mechanics and Maritime Sciences (M2)
Använd denna länk för att länka till samlingen:
Vi bedriver grundläggande och tillämpad forskning inom alla transportslag för att nå hållbara tekniklösningar. Forskning genomförs även för att bidra till miljövänlig processteknik och hållbar energiförsörjning.
Vid institutionen för mekanik och maritima vetenskaper utbildas framtidens ingenjörer och forskare med siktet inställt på övergången till ett hållbart transportsystem. Vår forskningsportfölj är unik i sin bredd och täcker in alla transportformer och verkar även för att bidra till miljövänlig processteknik och hållbar energiförsörjning. Genom samverkan med samhälle och näringsliv försöker vi lösa samhällets stora utmaningar - tillsammans.
För forskning och forskningspublikationer, se https://research.chalmers.se/organisation/mekanik-och-maritima-vetenskaper/
We carry out fundamental and applied research in all modes of transport to achieve sustainable technology solutions and to contribute to environmentally friendly process technology and sustainable energy supply.
At the Department of Mechanics and Maritime Sciences, the engineers and researchers of tomorrow are trained with their eyes set on the transition to a sustainable transport system. Our research portfolio is unique and covers all modes of transport and contributes to environmentally friendly process technology and sustainable energy supply. Through collaboration with the society and industry, we strive to solve society's major challenges - together.
Studying at the Department of Mechanics and Maritime Sciences at Chalmers
For research and research output, please visit https://research.chalmers.se/en/organization/mechanics-and-maritime-sciences/
Browse
Browsar Mekanik och maritima vetenskaper (M2) // Mechanics and Maritime Sciences (M2) efter Program "Computer science – algorithms, languages and logic (MPALG), MSc"
Sökresultat per sida
Sortera efter
- PostA natural language processing approach for identifying driving styles in curves(2016) McNabb, Eric; Kalander, Marcus; Chalmers tekniska högskola / Institutionen för tillämpad mekanik; Chalmers University of Technology / Department of Applied MechanicsA machine able to autonomously recognise driving styles has numerous applications, of which the most straightforward is to recognise risky behaviour. Such knowledge can be used to teach new drivers with the goal of reducing accidents in the future and increasing traffic safety for all road users. Furthermore, insurance companies can incentivise safe driving with lower premiums, which in turn can motivate a more careful driving style. Another application is within the field of autonomous vehicles where learning about driving styles is imperative for autonomous vehicles to be able to interact with other drivers in traffic. The first step towards identifying different driving styles is being able to recognise and distinguish between them. The aim of this thesis is to identify the indicators of aggressive driving in curves from a large amount of naturalistic driving data. The first step was finding curve sections to analyse within trips and the second step was reducing the data to become more manageable. Symbolic representations were used for the second preprocessing step, which in turn allowed the use of Natural Language Processing techniques for the analysis. We categorise drivers into different groups depending on their perceived tendency towards aggressive driving styles. This categorisation is used to compare the drivers and their driving style with each other. The tendencies used were Speeding, Braking, Jerky curve handling and Rough curve handling. Some general trends among the analysed drivers are also identified. It is possible to reuse the categorisation to include more drivers in the future or to use what we have learned about the features and drivers for further research.
- PostHighway tollgates traffic prediction using a stacked autoencoder neural network(2018) Kärrman, Oskar; Otterlind, Linnea; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Chalmers University of Technology / Department of Mechanics and Maritime SciencesTraffic flow prediction is an important area of research with a great number of applications such as route planning and congestion avoidance. This thesis explored artificial neural network performance as travel time and traffic volume predictors. Stacked autoencoder artificial neural networks were studied in particular due to recent promising performance in traffic flow prediction, and the result was compared to multilayer perceptron networks, a type of shallow artificial neural networks. The Taguchi design of experiments method was used to decide network parameters. Stacked autoencoder networks generally did not perform better than shallow networks, but the results indicated that a bigger dataset could favor stacked autoencoder networks. Using the Taguchi method did help cut down on number of experiments to test, but choosing network settings based on the Taguchi test results did not yield lower error than what was found during the Taguchi tests.
- PostHighway Tollgates Travel Time & Volume Predictions using Support Vector Regression with Scaling Methods(2017) Lin, Amanda Yan; Zhang, Mengcheng; Chalmers tekniska högskola / Institutionen för tillämpad mekanik; Chalmers University of Technology / Department of Applied MechanicsToll roads or controlled-access roads are widely used around the world, for instance in Asian countries. It is often expected that drivers can drive smoother and faster on the toll roads or controlled-access roads compared to on regular roads. However, long queues happen frequently on toll roads and cause lots of problems, especially at the tollgates. Accurate predictions of travel time and volume at the tollgates are necessary for traffic management authorities in order to take appropriate measures to control future traffic flow and to improve traffic safety. This thesis describes a novel investigation on the combination of Support Vector Regression (SVR) and scaling methods for highway tollgates travel time and volume predictions. The major contribution of this thesis includes 1) an approach to handling the missing data; 2) selection of important features; 3) investigation of three scaling methods and discussion of their suitability. Experiments were done as part of the Knowledge Discovery and Data Mining (KDD) Cup 2017.
- PostLane-Level Map Matching using Hidden Markov Models(2019) Korsberg, Ellen; Nordén, Eliza; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Chalmers University of Technology / Department of Mechanics and Maritime SciencesMap matching is the procedure of matching vehicle location and sensor data to a digital map. New high-definition maps, designed for autonomous vehicles, open up for the possibility of matching to lanes rather than roads. Inferring the lane-level positions of vehicles will be useful for updating and building probe-sourced maps, and thereby arguably essential for autonomous driving. This thesis seeks to solve the lane-level map matching problem using a Hidden Markov Model. The Viterbi algorithm is used to decode it. The model is tested on a data set yielded through the Volvo Drive Me project and collected by commercial vehicle sensors, including a GPS receiver, an inertial navigation system and a forward-looking camera. For the sake of simplicity, the RADAR and LiDAR sensors are excluded. Among the sensor data used, lane changes and the type of road lane markings as detected by the vehicle proves to be particularly important. Two metrics for evaluating model performance are proposed. The first metric is the recall, i.e. the fraction of correct matches. However, the lanes to which the observations are matched vary widely in length. Therefore, we introduce the path length error (PLE) as a complementary metric. As the name indicates, it considers the length of the incorrect routes. A naive matcher, that simply matches GPS coordinates to the closest lane, is used for benchmarking. Attaining 95% median recall and 3% median PLE, we conclude that our model is high-performing and robust to errors. For comparison, the naive matcher scores 77% median recall and 26% median PLE. Our model is however shown to struggle without reliable vision detections. It would therefore be meaningful to investigate the inclusion of additional vehicle sensors.
- PostSelf-organizing multi-agent systems for shared space operations: Using genetic algorithms and contract net protocols to solve the pickup and delivery problem(2018) Karlsson, Svante; Steffenburg, Jacob; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Chalmers University of Technology / Department of Mechanics and Maritime SciencesAutomation of on-site scheduling and route planning of units in a mining operation presents interesting challenges. The dynamic properties of real-life operations necessitate AI-inspired decentralized methods, which tend to be more robust under real conditions. Moreover, route planning in a mining environment, consisting of narrow passageways, requires vehicles to communicate and cooperate for safe and efficient transportation. We model this as a dynamic multiple-agent pickup and delivery problem; a problem in which multiple agents cooperate to complete transportation tasks that are revealed continuously. Taking inspiration from novel solutions, using auction-like bidding systems based on genetically optimized heuristics, we tackle the pickup and delivery problem (PDP) from two different angles. Firstly, we show that existing solutions to the planar variant of the PDP can be improved by giving agents the ability to communicate. Secondly, we present a solution method to the PDP in a shared space environment, applicable for real world scenarios such as a mining operation. Apart from using the aforementioned bidding system to assign tasks to agents, we also implement a method for solving decentralized multiple-agent path finding.
- PostTopic Modeling and Clustering for Analysis of Road Traffic Accidents(2017) Mekonnen, Agazi; Abdullayev, Shamsi; Chalmers tekniska högskola / Institutionen för tillämpad mekanik; Chalmers University of Technology / Department of Applied MechanicsIn this thesis, we examined different approaches on how to cluster, summarise and search accident descriptions in Swedish Traffic Accident Data Acquisition (STRADA) dataset. One of the central questions in this project was that how to retrieve similar documents if a query does not have any common words with relevant documents. Another question is how to increase similarity between documents which describe the same or similar scenarios in different words. We designed a new pre-processing technique using keyword extraction and word embeddings to address these issues. Theoretical and empirical results show the pre-processing technique employed improved the results of the examined topic modeling, clustering and document ranking methods.
- PostUsing Deep Neural Networks for Lane Change Identification(2019) Widjaja, Ryan Damarputra; Chalmers tekniska högskola / Institutionen för mekanik och maritima vetenskaper; Bärgman, Jonas; Streubel, Thomas; Kovaceva, JordankaLane change maneuvers are commonly performed by drivers in highway driving situations. It starts with the driver planning and making the decision whether a lane change maneuver is necessary according to the situation. Once the driver decides to do the maneuver, he/she starts to prepare themselves. Next, he/she changes their lateral position until the vehicle crosses the line between lanes. Finally after crossing the lane marking, the driver adapts to his/her new positions by stabilizing the vehicle. When lane change maneuvers are done incorrectly, accidents may happened which can be fatal to those involved in the crash. Advanced Driver Assistance Systems (ADAS) include several functions that rely on lane change detection e.g., lane departure warning (LDW) system and lane change assistance system. These functions can be used to help driver perform a safer lane change maneuver. Having a system which can accurately retrieve and recognize the driving characteristics of a lane change maneuver will be beneficial for the development of ADAS. Identifying lane changes in driving data can be done manually by annotations, but it costs a substantial amount of time and money in case of large driving databases. A cheaper solution would be to use machine learning algorithms as they excel in this type of problem. Several machine learning algorithms, specifically artificial neural networks, have been used in many different research applications, including lane change predictions. This thesis work included several steps. Firstly, driving data was retrieved from UDRIVE, which contains naturalistic driving data collected from various European countries. The data was processed into segments containing lane changes and baseline driving. It served as an input for training and testing using a sliding time window approach. Three neural networks were constructed to identify lane changes. One served as the baseline model, and the other two were variations of the baseline model called modified Long Short Term Memory (LSTM) and stacked LSTM, respectively. Training and testing were conducted to these networks using the same configuration and dataset. During the training process, some of the parameters were adjusted according to their performance and some could not be adjusted. Parameters that cannot be adjusted are called hyperparameters and they usually relate to the structure of a neural network model. Both the modified LSTM and stacked LSTM were subjected to parameter tuning, which is a process of changing various trainable parameter in order to make the model perform optimally. After parameter tuning was done, the best model was further evaluated by using cross validation. Results have shown that the stacked LSTM model has the best performance among the three models. It managed to reach F1 score of 0:7178 and able to identify 95% of the lane change data. However, the stacked LSTM model has performance problems in terms of training time in identifying the transition phases of lane change maneuver. Several factors which contributes to this performances issue are identified, such as the imbalanced training data, the variables selection, the structure of the network, the number of trainable parameters, the hyperparameter settings, and how the raw input data is processed. If these issues are resolved, it is expected that the stacked LSTM would have a higher performance in the range of 0:85 in F1 score.