Planning and development of software components (robotics, 2D/3D perception, backend, Android frontend) for the autonomous construction of apartment buildings using conventional construction cranes.

Concept and implementation of intelligent inventory management. Inventory measurement using LiDAR, automatic planning of storage location, recording location, and paths. PID control. Guidance of several junior staff members in the implementation of the concept and adaptation of existing software. Support for on-site commissioning in Finland.

Design and development of a processing pipeline for the detection of defects on optical systems using methods of classical image processing. The software enables fully autonomous evaluation even with high sensor noise.

Development and use of a training pipeline for semantic segmentation of timber loading cranes with additional weak annotation to reduce the number of full annotations required. The approach developed reduced the annotation costs of identical network topologies (U-Net) in the project by 40-50% while maintaining the same quality.

Software development for autonomous timber loading cranes in a 16-member team. Independent design and implementation of behavioral strategies and data structures for real-time execution in cognitive architectures. ?Hybrid? sensor processing using neural networks and classic image processing methods. Support for on-site commissioning in the USA.

Design and implementation of a software module for scanning documents with Android smartphones in real time using classical computer vision.