Master’s Thesis: Intelligent sensor platform modelling using AI
About the company
ÅF is an engineering and design company within the fields of energy, industry and infrastructure. We create sustainable solutions for the next generation through talented people and technology. We are based in Europe and our business and clients are found all over the world.
ÅF Making Future.
About the job
ÅF Embedded Systems West has approximately 180 employees. We are a design and development house for advanced electronics and software systems. We run numerous development projects for industrial, automotive, military and medtech customers.
ÅF Embedded runs master’s thesis on regular basis and we now seek skilled students to run this master’s thesis project in a new exiting area where the combination of the latest sensing technology merges with the latest AI technology in an embedded edge computing system.
If you succeed with this challenge, you may be one of our next development engineers.
You, together with ÅF, will develop a new low end sensing platform that can sense objects in a 360 view at a maximum distance of 5 meters. The platform aims to enable products that are to be a low cost alternative to 360 LIDAR for medium range (5 meters) applications.
This will be done by bringing multiple low cost state of the art Time of Flight (ToF) range sensors together as a system by using AI implemented in a low cost FPGA.
The master’s thesis will be performed in simulated environments and will primarily involve software tools and products.
Proposed thesis tasks
There are some manufactures of medium range ToF sensors that has a broad field of view.A sensor model that can be used to generate training dataset input to the network shall be developed for at least one physical sensor (e.g. ST or Renesas). The sensor model must at least contain range data (object distance), data rate, and the critical configurable settings. This can be done in e.g. Matlab.
Sensor system modelling
Multiple (approx. 10-20) sensor are planned to be mounted in a ring topology. In order to simulate system sensor data input, a sensor system model needs to be developed based on the individual sensor models. It can be assumed that the sensor data is synchronized in time.
Neural network design
The neural network shall use simulated input from the sensor system model. The network shall output the distance and angle to one or more objects. The design shall be simulated by using the Lattice software toolchain.
Simulated training dataset input
Training data needs to be generated in order to train the neural network to behave correctly. This data generation may require a simple external object model and a simulation.
Simulated use case data input
To test the performance of the network, simulated use cases needs to be defined and simulated dataset needs to be generated.
Network feasibility and verification
A full system simulation based on the use case data input needs to be executed to verify if the solution is feasible.
Who are you?
Candidates’ background and skills
Master of Science (30 credits) in Electronics / Computer / Mechatronics or equivalent. Students with an interest in intelligent sensor systems, software development, Machine Learning / AI and data-processing / analysis are welcome to apply. Good knowledge in programming is a requirement. Big plus if you have programming and software development as a hobby interest, as well as some working experience from previous job/summer job or other relevant experience. As a person, you are self-going and self-motivated with an outgoing can-do-attitude. Join ÅF during your master thesis and experience an environment of brave, devoted and highly skilled team players.
Place of work will be at the ÅF office in Gothenburg. Supervision is carried out by ÅF employee at the office in Gothenburg in collaboration with the supervisor at the university. The starting date of work is flexible.
How to apply
We prefer that the thesis is performed in groups of two. You and your thesis partner apply individually with a personal resume and a joint cover letter. Use your joint cover letter for both of your individual applications.
Anna Nilsson, recruiter