Four thesis projects available at the Autonomous Mobile Manipulation lab
The Automonous Mobile Manipulation lab (https://amm.aass.oru.se/) at Orebro University is offering four thesis projects suitable for an MSc or civ.ing. 20 week thesis project. In exceptional cases, a team of 2 Bsc or hing students can also be considered. Please get in touch with todor.stoyanov@oru.se to apply. Include a CV, motivation letter (1-page limit), and grades transcript.
Cable routing and assembly
Modern vehicles have a large number of electronics components, which all rely on a large number of cables for power and connectivity. Assembling these cables during vehicle manufacture is still done as a largely manual task, which is not always ergonomic and requires a lot of human effort. In this project your task will be to investigate the use of a collaborative robot for routing and connecting electric cables in a mock-up vehicle module. You will examine how a human demonstrating the task can transfer knowledge and instructions to a robot, with a main challenge of deducing a sequence of intermediate goals for the robot to follow in order to achieve the assembly. This project is given in collaboration with Volvo GTO.
Expected background: Excellent programming skills (C++ and Python), excellent background knowledge in robotics, experience in sensors and robot programming is a plus.
Sensors for Localization in Underground Mines
Above ground it is common to use GPS in order to estimate rough vehicle position and orientation and aid in fine-grained localization of autonomous systems. This option is, however, not available for systems operating in underground mines and tunnels, severely limiting autonomy. In addition, underground mines in particular are difficult environments for sensors, as they are dark, poor in features, and often include dust and water vapor that obstruct sensor views. In this project your task will be to evaluate and compare a number of different sensing technologies with potential application for localization in underground mines. This project includes a strong field-work component, hardware integration, data collection in a real mining environment, and data processing and analysis. This project includes work within a Vinnova collaborative research projects together with Epiroc and Boliden.
Expected background: Good theoretical knowledge of different sensing technologies, comfortable with Linux and embedded systems, affinity for field robotics work.
Learning Features for Underground Localization (ORU)
Above ground it is common to use GPS in order to estimate rough vehicle position and orientation and aid in fine-grained localization of autonomous systems. This option is, however, not available for systems operating in underground mines and tunnels, severely limiting autonomy. In this project, you will investigate methods for feature-based localization in underground mines. You will begin by collecting a large-scale simulated data set with ground-truth localization and realistic sensor models. You will then train feature detectors and investigate feature-based localization methods in your simulated data set. You will collaborate with another student to collect a real data set and then evaluate your algorithms also in the real data. This project includes work within a Vinnova collaborative research projects together with Epiroc and Boliden.
Expected background: Good theoretical background in machine learning and image processing, excellent programming skills (C++ and Python), background on sensor technology and simulation are plus.
Realistic Range Sensor Simulation
Simulators are important for developing and prototyping robotics algorithms, however, accurate physics simulation is usually difficult for current open-source tools. An alternative is to use a commercial physics engine, which has many advantages, but one notable problem: available tools do not readily simulate the relevant robotics sensors. In this project, you will contribute to an open-source effort to build a robotics simulator around a commercial (closed source) physics engine. Your task will be to focus on simulating sensors, and range sensor in particular, with an accent on efficient implementation, parallelization, and realistic sensor noise models. This project includes collaboration with our industrial partners at Algoryx.
Expected background: Excellent programming skills in C++, good theoretical background on sensors, prior experience with open source frameworks and ROS desirable, and experience with parallel programming and CUDA is a plus.
Annonsuppgifter
Annonsör: Örebro universitet
Ansök senast:
Annonskategori: Examensarbete, praktik, uppsats
Intresseområde: Data och IT
Kontaktperson: Todor Stoyanov (Associate Professor) todor.stoyanov@oru.se
Webbsida: https://www.oru.se/english/research/research-environments/ent/aass/