Processing lidAR data from a virtual logistics space

We study computing solutions that can be used close to the network edge in I2oT systems (Industrial Internet of Things). As a specific use case, we consider a factory warehouse with AGVs (Automated Guided Vehicles). The computing services for such systems should be dependable, yield high performance, and have low latency. For understanding such systems, we have constructed a hybrid system that consists of a simulator yielding virtual LiDAR sensor data streams in real-time and a sensor data processor on a real cluster that acts as a fog computing node close to the warehouse. The processing merges the observations done from the individual sensor streams without using the vehicle-to-vehicle communication links for the complicated computing. We present our experimental results, which show the feasibility of the computing solution.