Machine learning for dynamic obstacle predictions
Dnr:
NAISS 2025/22-548
Type:
NAISS Small Compute
Principal Investigator:
Knut Åkesson
Affiliation:
Chalmers tekniska högskola
Start Date:
2025-04-04
End Date:
2026-05-01
Primary Classification:
20201: Robotics and automation
Webpage:
Allocation
Abstract
Given seqeuneces of images captured by cameras we are aiming at doing semantic segementation using semi-supervised learning together with forecasting of motion patterns of dynamic objects. The overall goal is to to use the information to control a fleet of autonomous transport robots in a factory. This work is used in the project AiMCoR and SECAR funded by CHAIR, WASP and AB Volvo.