Edge-Enabled Autonomous Navigation and Computer Vision as a Service: A Study on Mobile Robot’s Onboard Energy Consumption and Computing Requirements

Abstract

Autonomous mobile transport systems are a crucial part of flexible factory logistics enabling a highly-adaptable industrial production organization. Edge computing bares the potential to support commissioning and operation of these mobile robots through offloading computation in a very flexible and scalable manner. Among these benefits, less onboard computation means also less energy consumption and longer operation durations. In order to evaluate the specific effects of offloading computing following the Software as a Service principle, we consider different software service distribution scenarios in terms of real-time critical navigation as well as additional computer vision functions. We present an empirically study using and assessing 4G radio networks in order to realize distributed control scenarios within industrial environments. The software services are offloaded to the edge following a modern microservice approach. Our results show that onboard computing can be covered by a low-cost single-board computer and energy consumption can be reduced by 4.9% to 18.4% through offloading computation depending on the payload and average velocity. Regarding to additional computer vision algorithms, the energy savings are even higher.