AWS for Industries

Makino improves performance of Autonomous Mobile Robots with AWS Wavelength and 5G network


Makino Milling Machine Co. (Makino), a Japanese milling machine manufacturer, has successfully launched a robot control system over a 5G network in less than 5 months. Makino has improved the stability of wireless communication between its moving Autonomous Mobile Robot (AMR) and the control server using the 5G network and Amazon Web Services (AWS). The company opted for Public 5G and AWS Wavelength, which resulted in significant cost savings compared to building a Private 5G environment themselves. This blog describes the challenges Makino faced and the solutions it implemented.

Robots working in factory

A device commonly known as a machining center requires the replacement of heavy tools, such as milling cutters and bits, to perform various metal processing tasks. Currently, these tools are manually replaced by human workers, which involves physically demanding labor. Makino manufactures and sells their own AMR, which is designed to automate the transportation and replacement of tools within factories. Makino’s AMR is equipped with LiDAR sensors, allowing it to detect surrounding obstacles and move autonomously. It is designed to collaborate with workers and other AMRs on the shop floor.

Figure 1: Makino’s Autonomous Mobile Robot

Figure 2: Makino’s AMR can recognize its surroundings with LiDAR

Figure 3: Makino’s AMR Dashboard displays Auto Mapping using LiDAR


Makino’s AMR receives orders from the control server and can move autonomously to fulfill these orders. Previously, Makino used Wi-Fi for communication between the AMRs and the server. The server transmitted order messages to the AMR using several protocols, such as HTTPS and WebRTC, which required stable and uninterrupted communication. However, Makino has frequently experienced instability in the communication between the moving AMR and the server. When the AMR moves out of range of a Wi-Fi access point and loses connection, the handover to another access point takes a significant amount of time. Makino has identified that the delay in the Wi-Fi handover process is the cause of this instability. Therefore, a stable handover to another access point is required.

Figure 4: Makino’s challenges

5G realizes stable wireless communication for autonomous mobile robot applications

Makino has adopted a 5G network instead of Wi-Fi. 5G communication is designed to provide seamless connectivity while on the move, allowing for smooth handover between base stations even when devices are in motion.

There are several ways to achieve 5G communication indoors in factories. For example, factory owners may build their own Private 5G networks, referred to as “Local 5G”. Makino conducted an evaluation of different wireless communication methods, specifically comparing Private 5G with Public 5G.

Meanwhile, Makino has opted for Public 5G. As a result, Makino has achieved both cost reductions and real-time stable operability. Along with adopting a 5G network, Makino has migrated the servers that were previously in the factories to AWS Wavelength, a hybrid cloud service using Public 5G, as a focal point of the research.

Why Makino chose Public 5G and AWS Wavelength

AWS Wavelength is an AWS service that performs processing with AWS resources on a 5G base station network owned by a mobile carrier. With its capability to perform processing and respond in close proximity to the wireless base station, AWS Wavelength is well-suited for low-latency processing in 5G communication. KDDI, a major mobile carrier in Japan, supports AWS Wavelength.

Figure 5: AWS Wavelength

Makino concluded that the combination of Public 5G and AWS Wavelength is a better option for its use case compared to Private 5G because of these benefits:

Network stability and performance

Firstly, Makino praised the excellent performance of 5G communication in mobile connectivity. The company installed 5G base stations in its factories and evaluated network stability and performance. Makino has confirmed that there were no communication disconnections even when a moving object communicated across multiple 5G base station areas. Additionally, Makino measured the performance of communication. The network latency from the instance on the Wavelength Zone to the AMR was measured to be between 10 and 15 milliseconds, and the overall latency from the control application to the AMR was approximately 40 milliseconds, measured using an application that utilized HTTPS over a VPN tunnel. This latency is short enough to send messages to control Makino’s AMR. Currently, Makino has adopted a “lift” approach, prioritizing the time to launch over the effects of application optimization. The communication architecture used on premises still operates on the Wavelength Zone. Optimizing the application for AWS Wavelength may potentially lead to lower latency in the future. The network throughput was measured at an average of 1.1 Gbps downstream and an average of 140 Mbps upstream, which was a satisfying result for Makino.

Lower initial costs

Makino placed particular importance on costs in their decision making. Public 5G networks have the advantage of not requiring expenses for 5G equipment investment and maintenance, unlike operating Private 5G networks.


There is no need to obtain a license for operating a Private 5G communication network or to employ experts. Public 5G can be used with a general smartphone. While 5G carriers continuously upgrade base stations, owners of local 5G networks are responsible for upgrading and repairing their own equipment.

Monitoring and investigating robots remotely

Previously, AMRs were monitored and managed within the factories. The control servers were located in the factories and connected to the factories’ LAN, making external management unfeasible. However, with the utilization of Public 5G, Makino is now capable of remote management, enabling remote investigation of AMR operations. In the event of an issue with AMR, an operator can remotely investigate and take appropriate action. Additionally, more detailed logs can now be viewed remotely in real time.

No need to place any servers in factory

Makino has succeeded in eliminating the need for control servers to be installed together with AMRs in the factories by using AWS Wavelength. Consequently, customers purchasing robots from Makino are no longer required to set up a server infrastructure within their own factories. This simplifies the process for Makino customers to purchase advanced robots.

Network configuration with AWS Wavelength

Figure 6: AWS architecture diagram of the solution

The control server of AMRs is located in the Wavelength Zone. AMRs and the server communicate through the Public 5G network. The tablet for control and the server communicate over the internet.

Makino took several factors into consideration when using AWS Wavelength.

IP address range: Generally, factory equipment often operates using private IP addresses. Makino’s AMR also required a specific private network range. On the other hand, the range of IP addresses attached by AWS Wavelength consists of carrier-specific global IP addresses.

Device authentication: AWS Wavelength does not support a managed service for AMR device authentication, such as AWS IoT Core.

Figure 7: Consideration points for using Makino’s AMR with AWS Wavelength and Public 5G network

After careful consideration, Makino has found a solution to solve these issues. It created a VPN router instance in the Wavelength Zone. The VPN router establishes VPN tunnels to each AMR, enabling communication from the private IP address range of AMRs. Device authentication is performed using a device certificate during the creation of the VPN tunnel.

Figure 8: Solution to use Makino’s AMR with AWS Wavelength and Public 5G network

As a result, Makino was able to securely communicate with its AMRs and servers using AWS Wavelength.


In less than just 5 months using a Public 5G network and AWS Wavelength, Makino has achieved the following benefits:

  • Network stability and network performance
  • Lower initial cost
  • Maintainability
  • Monitoring and investigating robots remotely
  • Serverless AMR system in factory

Makino plans to sell this optimized system to its customers to enhance their experience. Makino recognized the possibility of further operational improvements and improved network performance by optimizing architectures using AWS, making them faster and easier to maintain and operate.

Learn more

Watch Makino’s story from an AWS Japan event in April of 2022, where they explain their autonomous mobile robot and this use case with 5G network and AWS Wavelength. These slides and videos are in Japanese.

Kohei Yoshikawa

Kohei Yoshikawa

Kohei Yoshikawa is a senior solutions architect at AWS Japan. After graduating from the master’s program at Hokkaido University, he worked as a software developer and system integrator for over twenty years. He joined AWS in December 2020 and has been supporting many manufacturing customers in Japan. He enjoys cycling on the weekends and skiing in the winter.