Machine learning in robotics and edge devices for space exploration

The course provides a wide insight into neural network (NN) algorithms and their hardware implementation. The development of NN applications is typically carried out using GPUs and requires a long calculation time. The course gives participants the ability to accelerate and shorten inferring latency using dedicated hardware with limited resources. Although this idea can be adopted in many different applications from many disciplines, the course demonstrates practical examples from space-related research projects. The main goal is to motivate, illustrate, and experience the impact of Machine Language (ML) and Artificial Intelligence (AI) on the space sector.

Besides the emphasis on creating practical design on the available hardware platforms, the course presents a survey of commercially available (and recently introduced by leading manufacturers) systems for hardware implementation of neural algorithms. A survey of dedicated processors with neural architectures currently being developed is also covered.

Another fascinating subject in the course concerns emerging technologies dedicated to future hardware neuromorphic systems currently in the R&D stage.

The huge innovation potential of ML is strongly emphasised during the course. Practical examples of innovative projects are presented to illustrate the impact of ML and AI in business activity.