Design of a human-machine-interface for integration with an automated robotic vehicle

Abstract

Automated Guided Vehicles (AGVs) and Automated Robotic Vehicles (ARVs) are increasingly being integrated into manufacturing factories. However, a prevalent challenge that arises in areas where both these vehicles and human workers operate is the frequent interruption of vehicle movement due to individuals inadvertently obstructing their path. The aim of this dissertation was to come up with a system that prevents such interactions at STMicroelectronics. Following a visit to STMicroelectronics and consultations with the engineering team, it was determined that a floor projection system would offer the most effective solution. Concurrently, research was conducted on AGV/ARVs, collaborative robots, and analogous solutions to inform the development process. Customer requirements were analysed and a house of quality exercise to identify the optimal solution was conducted. This concluded that the AGV-mounted projector system was best, with a Roof-mounted system developed for comparison. Through evaluation of various sensory technologies, a machine vision human detection algorithm was chosen for its ability to distinguish humans from other objects. This algorithm was integrated with a basic Human Machine Interface, projecting a warning, and emitting a sound. The control system chosen for this project was the Raspberry Pi 4 Model B. Three code iterations were developed which monitor people from Face Distance measurement, Face or Pose detection and Pose only detection. Both prototype systems underwent testing, with the AGV-mounted system tested on a trolley at STMicroelectronics and the Roof-mounted system tested at the University of Malta. Despite marginal performance from both systems, the AGV-mounted system showed slight improvement over the Roof-mounted one in both iteration time and catching people attention. The limited performance is mainly attributed to projector limitations and human adaptation. The Roof-mounted system's performance was also affected by projector constraints and testing environment factors. The analysis favours the AGV-mounted system over the Roof-mounted system for further development. Improvements are needed in both hardware and software.