5G-EVE: Industry 4.0: Autonomous vehicles in manufacturing environments

Overview

The objective of this use case is to assess the viability of operating in factories 5G connected Autonomous Guided Vehicles (AGVs), with the control of the vehicle virtualized at the edge of the network i.e., moving the control of the vehicle out of the physical unit and implementing it in a computing node that meets the latency requirements for the AGV operation. The AGV collects the information from its sensors and sends it to the virtual controller through a wireless connection. This information from the sensors is processed in the virtual controller, identified as virtual PLC, which generates the orders to be executed by the AGV actuators. These orders are sent again through the wireless connection to the AGV, where they lead the actions of the different actuators in the next action period.

Architecture

The trial architecture encompasses: 

• Radio Access network supporting high performance 5G NSA access.
• Virtual EPC supporting NSA access.
• Virtualized processing platforms, implementing the functionalities required to support the trial, virtual PLC and real time video image recognition.
• Orchestration platform, in charge of instantiating the processing functions, as well as configuring the VNFs.
• Measurement infrastructure based con Kafka bus, that collects measurements from network and processing functions to derive network and service KPIs.

The system is connected to the Interworking Layer (IWL) in Turin to allow the launch of tests from the 5G EVE portal.

Deployments

The trial has been implemented at 5TONIC lab (https://www.5tonic.org), where a circuit for two AGVs has been deployed.

5G NSA coverage is provided at the lab, operating in bands 7 and n78.

The AGVs incorporate two wireless routers that support two connections linked to:

• Implementation of the virtual PLC controlling the AGV, in charge of different procedures:
  
  • Tracking the marked route.
  • Collision avoidance of obstacles in the AGV route.
  • Actions associated with tracking marks deployed in the route (e.g., stopping the AGV for loading)
• Real time image processing, mainly to identify what obstacle (person, object, etc.) has activated the collision avoidance mechanism. It is supported by a camera deployed in the AGV.

The controlled AGV is an Easybot model, designed and manufactured by ASTI and used in real factories worldwide.

Results

The trial demonstrated that it may be feasible to support an operating model based on the use of virtual PLCs with 5G to control a fleet of AGVs operating in a factory. Measured latency was 10 ms, which is in the lower bound of ASTI’s proposed requirements, and reliability exceeded 99,9999%.

Supporting this operational model has several potential advantages for current and future use of AGVs in factories:

• Potential coordination among AGVs, optimization of routes and rerouting in case of failure, are all facilitated implementing a centralized control of the factory’s AGVs.
• Lower cost of the AGV, as it has not to incorporate the processing capabilities (which would be critical for the future, as otherwise AGVs would be required to generate maps from LIDAR measurements).
• Easier maintenance of the equipment, e.g., software upgrades have not to be carried out for each AGV.
• Higher reliability through the virtualization of the centralized processing.

The trial was demonstrated at EuCNC 2019, and has been shown to verticals such as PSA, Volkswagen, Innovalia or Gestamp Group.

5G Empowerment

The operational model, for being viable in a real factory, requires that the performance of the AGVs operation remains the same as if the classical, on-board control would be maintained. ASTI estimates that the following network related KPIs should be achieved:

• Latency: to ensure that the AGV works correctly in all operating conditions, the maximum delay (between the generation of the measurements by the sensors till the associated orders are received at the AGV) should not be greater than 10-15 ms.
• Reliability: 99.999% probability of correct reception of packets in both directions of the link.

Throughput and capacity are not critical parameters for this use case, although they may become in its evolution. Also a zero latency handover would be quite beneficial.

All these KPIs cannot be achieved using 4G technology (although suboptimal operation or operating in conditions that does not require the KPIs indicated above is feasible with 4G and even Wi-Fi), so 5G becomes an enabler for the support of control of AGV fleets with virtual PLCs.

Trial Dates: Q2-2019

.https://www.5g-eve.eu/