Safer Autonomous Block Stacking in Human Shared Warehouses : Using Sensors as Replacement For a Mechanical Safety Function

Global trade has undergone rapid expansion in recent decades, in many ways due to the rise of e-commerce. This has led to increased globalization and increasingly complex supply chains. This in turn introduces a large demand for increasingly efficient logistics, driving the shift towards the automatization of said logistics. Toyota Material Handling in Sweden produces forklifts for warehouse use, with autonomous models emerging as the clear path for the future. With these new autonomous forklifts working alongside humans, safety is paramount. The aim of this thesis is to see if a mechanical fork-tip safety bumper can be replaced by sensors in the specific case of block stacking pallets. This could be achieved either by detecting humans or other obstacles in the upcoming path by means of some sensor, or alternatively replacing the bumper with a short range virtual bumper. To evaluate if these solutions were practically feasible, a theoretical study was performed followed by some experiments to test possible solutions. Given the industrial context of the project, solutions did not only have to be technically viable. Any feasible solution also had to meet safety and cost-effectiveness requirements, which proved to be a significant challenge. The findings indicate that real-time obstacle detection using the current sensor infrastructure and environmental awareness of the autonomous forklifts is not practically achievable in the case of tight block stacking. However, the virtual bumper approach was found to be very promising. Furthermore, the thesis proposes the integration of cameras in the fork tips as a promising complementary solution for enhanced object detection and safety.