Fusion multicapteurs pour une navigation sûre des robots mobiles autonomes dans les entrepôts logistiques

In Logistics 4.0, handling tasks are being robotised. The introduction of mobile robots is a first step towards replacing human for transport and avoiding the time wasted wandering through aisles. However, these mobile robots, which share the same geographical space as human operators, must respect certain constraints to ensure the physical and psychological safety of the operators. The robot, now becoming a cobot, must learn to distinguish between human and non-human obstacles. This paper focuses on this issue. We show that 2D LIDARs information is not sufficient to manage these issues and that this information must be merged with data from other sensors such as RGB-D cameras to ensure that the cobot's behaviour is safe and socially acceptable.