Research

Drone swarm flight

What if controlling one drone were enough to fly a whole swarm?

The research involved one controlled drone and three others that were intended to position themselves based on its location and each other's locations.

The idea was that the pilot could control one drone while the others followed it, maintaining their formation. This would let one person control several drones at the same time.

Such a system could be useful in searches for missing people, for example. Several drones could search an area at once without needing a separate pilot for each one.

I implemented a Python-based control system that sent position data from four drones to a server. The system directed three follower drones into formation.

Identifying road surface conditions in infrared images

Could observations that require expensive specialist equipment also be made with a computer costing a few tens of euros and an inexpensive camera?

There are expensive specialist systems for identifying road surface conditions. They can detect things such as ice forming on the road.

The research explored which similar observations might be possible using a Raspberry Pi and an inexpensive camera.

What particularly interested me was how far simple, inexpensive technology could go when used in the right way.

I implemented image recognition software that classified the ground surface as dry, wet, snowy or icy.

Can technical analysis predict cryptocurrency markets?

Status: New research about to begin

I'm investigating how well technical analysis works for both established and smaller cryptocurrencies. I want to find out whether apparent successes come from a consistent edge or chance.

The plan is to test methods on historical market data, account for trading costs and investigate whether the patterns hold up on data that wasn't used to develop the methods.

The aim is to turn the research into a scientific publication.

Can AI find something in the markets that traditional methods miss?

Status: Research idea

I want to compare machine learning methods with traditional analytical methods and find out whether they produce better predictions.

I'm particularly interested in whether any advantage holds up when market conditions change.

Could quantum machine learning help with market analysis?

Status: Research idea

Quantum computing and quantum machine learning interest me especially because there is still so much to find out about their actual practical benefits.

I'd like to investigate whether there are problems where quantum methods could offer an advantage over classical computational methods, and how any benefit could be demonstrated reliably.