Bilal Tariq, Lukas Wiesmeier, Karl Schulz
Minor rubbish left in grass and gravel areas is an ecological and economic problem. Since conventional sweeping machines designed for flat asphalted surfaces cannot be used, these areas still have to be cleaned manually. Angsa fundamentally changes the way grass and gravel are cleaned. Its core innovation lies in an artificial neural network architecture that can detect and localise individual objects based on video recordings. The individual objects are then removed selectively using a special suction technique. Angsa thus not only reduces the negative impact of rubbish on the environment; it also creates an economic benefit for its customers. The Angsa solution targets a variety of use cases, including the cleaning of festival venues after events and the daily cleaning of parks and other green spaces. Since the robot operates on open terrain using a few keypoints for orientation, localisation via GNSS is essential to its operation. The precision of this system is highly relevant to the robot’s area coverage: With better localisation, the robot can plan a more efficient path and clean a given area more quickly.