Master thesis proposals - external

Ericsson Research Looking for a thesis project during FallSpring, starting in SeptemberJanuary 20234? Did you know Ericsson Research develops research on themes related to robotics and AI? Take a look at the current projects we have open:

Project 1: Path planning for UAVs using the Digital Airspace¶ Ericsson Research is looking for a student to carry out a MSc Thesis project during Fall 2023 on path planning for UAVs connected to a UTM (Unmanned aircraft system traffic management) through cellular network. The main goal is to develop a path planning algorithm that accounts for the challenges and risks encountered when operating over a city. Applicants should have a solid background on planning algorithms, programming skills (Python, C++), as well as strong communication skills in written and spoken English. Ericsson Research - Short MSc Thesis Description - Spring 2024

Requirements

All the projects require some common skills:

* Strong ability to formulate problems and solve them, independently and in groups.
* Programming skills, preferably in Python or C/C++.
* Strong communication skills in written and spoken English.
To apply or get more information

Project 1: Fernando dos Santos Barbosa - fernando.dos.santos.barbosa@ericsson.com

ABB Robotics Automatic Learning of Behavior Trees for Industrial Robotics Today, industrial robots can solve very complex tasks in controlled environments, but modern industrial applications in workspaces shared with humans, require robots to be able to operate in unpredictable surroundings as well. One common way to deal with this is to control the robot with a reactive policy such as Behavior Trees (BTs). Also, with increased variation and smaller production batches, the time needed for programming, system integration, validation and verification is prohibitively long. It is therefore desirable that new robot policies or programs can be created automatically.

One way to generate policies automatically is machine learning algorithms. A drawback is that many of the most efficient machine learning algorithms, often based on Reinforcement Learning, are adapted to neural networks that have several important disadvantages compared to BTs and other model representations, as neural networks are not particularly transparent nor as modular.

Start: January 2023Duration: 6 monthsPlace: ABB Robotics (Västerås)ABB will cover the accommodation in VästeråsContact: Jonathan StyrudDetails here.

SCANIA Smart Factory Lab Scania’s Global Industrial Development has a Smart Factory Lab Team where new techniques are adapted, evaluated, demonstrated and implemented. Within this area we have a number of possible topic proposals:

* AI for Human Well-Being ( Human Pose Estimation / Motion Capture)
* AI for Computer vision (Generate synthetics training data for production/Object detection for quality inspection)
* Two autonomous mobile robots exchange material

* Modern navigation and planning of mobile robots (with ROS)
* AI for improving Energy Consumption and System Performance in Production
We are looking for a number of students to collaborate with us. Do you have another idea? Projects may also be initiated by you, so we welcome suggestions. Please indicate your interest area in the cover letter.

Contact person: Xiaomeng Zhu (xiaomeng.zhu@scania.com)

More details and Apple here: 30 credits – Smart Factory Lab - Machine Learning and Robotics for Future Industrial Digitalization and Automation

Hitachi Energy The electrical power system is evolving due to e.g. an increasing amount of renewables and distributed power generation. FACTS (Flexible AC Transmission Systems) devices can assist with this evolution by improving the power quality and reducing the risk of system disconnections.

In this master thesis proposal, we want to investigate the application of machine learning to a FACTS device. The device itself is based on a power electronic converter and has its own local controller. The performance, in terms of efficiency and stability, is dependent on modelling assumptions which may to a certain extent deviate from reality. Here, the objective would be to improve the performance of the FACTS device, by application and integration of a suitable machine learning method into the control. The master thesis is planned to start in January 2023.

Contact: Jonathan Hanning, jonathan.hanning@hitachienergy.com

For more details, see: https://www.hitachienergy.com/career/jobs/details/SE53908798_E1

Ocean Infinitiy Ocean Infinity has a number of proposals for thesis projects during Fall, starting in the end of August 2022. These involve problems within underwater perception, sonar and robotics. For example:

* Target Identification and classification from sidescan sonar data.
* Marine Growth assessment using computer vision, (Marine Growth Classification)
* MultiBeam Echo Sounder (MBES) image alignment (Multibeam Sonar Image Alignment)
* Image quality assessment (visibility, exposure, lighting, positioning etc) (Automated Image Assessment)
* Localization of AUV swarm (SLAM loop closing)
Please contact John Folkessn johnf@kth.se for more information.

Arriver Sweden Knowledge Distillation for Autonomous Driving

Description

CNN-based association for Multiple Object Tracking

Description

Univrses Cross-city domain shift is the cause of a 25-30% metrics drop of normal Deep Learning predictors. Change of weather and lighting conditions might even result in more disruptive effects. Therefore, to provide a robust and accurate output, the predictor would need to be trained on every possible deployment condition, e.g. multiple cites, districts, seasons, kinds of weather, cameras pose and intrinsics. This largely reduces the scalability of AI solutions both in terms of costs and deployment time. Recent Unsupervised Domain Adaptation (UDA) methods proved to be very effective in mitigating (or even solving) these shortcomings.

Contact: Pier Luigi Dovesi <pier.luigi@univrses.com>Application link, full description and details: https://career.univrses.com/jobs/1401972-master-thesis-project-unsupervised-domain-adaptation

ABB Corporate Research Using dialogue to disambiguate robot manipulation, learning from demonstration.Behavior Trees are a reactive task switching policy representation, used to control robotic agents. LfD can be used to teach the robot a task and BTs can be generated out of human demonstrations. However, ambiguities might rise if the target object for the task is similar to other objects in the environment. Thus, verbal-HRI can be used to disambiguate the task, both during the learning step and the execution step. The objective of the project is then to build a framework for continuous interaction between human and robot, from learning a task to execute it.Start: between Jan. 2022 and June 2022Duration: 6 monthsPlace: ABB CRC (Västerås)ABB will cover the accommodation in VästeråsContact: Matteo Iovino, matteo.Iovino@se.abb.comDetails here.

SkyCraft AB Master projects in CV and ML on powerline detection

Description

Viscando AB Projects within deep learning, signal processing and modelling for traffic and autonomous vehicle safety

Description

Babyshop Group Leveraging transfer learning for multi-label attribute prediction in fashion images, using deep learning

At Babyshop Group, there exist tens of thousands of images of children’s apparel. Currently, new items are labeled by hand, where each item has multiple attributes, eg. color, category, pattern, neckline, and style, etc. Since the workload of this manual task is immense, a need for a support system is high. In literature, a multitude of previous work has been done in regards to multi-label attribute prediction on fashion clothing, most notably DeepFashion and iMaterialist Fashion. Here, deep learning models have been trained on hundreds of thousands of labeled images to predict corresponding attributes, which has proven highly successful. However, almost all of these images are of adult fashion, with very little emphasis on childrens’ fashion. The question then arises, is it possible to combine the pre-trained models from previously mentioned multi-label attribution prediction models that are trained on almost exclusively adults’ fashion, together with the learnings from transfer-learning frameworks - in order to successfully perform multi-label attribution prediction on childrens’ fashion?

Required qualifications: MSc studies in Computer Science, Machine Learning, Computer Engineering, Mathematics, Physics, or related field; Good understanding of machine learning frameworks such as Keras, TensorFlow, PyTorch, Scikit-Learn, and/or Spark; Proficiency in Python and Git; Knowledge about data wrangling and data munging, using SQL, Pandas, and Numpy.

Web version of the proposal with instructions on how to apply.

Time frame: from early January to mid-June.

Contact: Marcus Svensson (marcus.svensson@babyshop.se), Data Scientist at Babyshop Group