Mobile robotic systems such as Unmanned Ground Vehicles, Unmanned Surface Vehicles, Unmanned Aerial Vehicles, and Autonomous Underwater Vehicles are becoming increasingly prevalent today with a broad number of applications including defense, medical, logistics, manufacturing support, and personal use.
With the rise of Artificial Intelligence and Machine Learning, many of these systems can function in autonomous or semi-autonomous states. To get some perspective on how intrinsically this will be tied to our lives over time, it is beneficial to gain some perspective on how fast the market is growing.
If the entire scope of applications is considered, some estimates size the market at $54.1B by 2024 ($18.7B 2018) which amounts to a 23.71% CAGR. If the focus is narrowed to Autonomous Guided Vehicles (AGV) used in logistics and manufacturing support, the market is expected to reach around $12B by 2025 with an estimated 15.8% CAGR.
This is based on a rough averaging of multiple sources as data is somewhat variable. The key takeaway when looking at the market growth for these devices is that they are going to continue to be a significant and increasing part of our manufacturing and logistics environment.
From a connectivity standpoint, these systems rely heavily on wireless networks and as autonomous vehicles are increasingly used, these networks will need to be scalable and reliable enough to meet these needs.
Wireless Networks for Autonomous Vehicles-Today
The most commonly used technology for wireless connectivity to mobile robotic systems is based on the IEEE 802.11x set of standards commonly known as Wi-Fi™. These networks transmit at publicly available frequencies within either the 2.4 GHz or 5 GHz space and are sliced into channels usually 20 MHz or 40 MHz wide (adjacent channels bonded).
These channels are essentially the pipes that data can pass through and the width of these channels significantly impacts throughput in the system. Over time, the IEEE 802.11x specification has evolved to incorporate changes in areas such as radio modulation and antenna structure (MIMO) to improve overall system throughput and reliability.
Updated standards such as 802.11n and 802.11ac can realize networks' speeds of several hundred bits per second 802.11x-based networks are based on the lower layer Ethernet protocols for their network structure with both its use of MAC addressing for unique device identifiers and IP protocols for logical network addressing. This gives a rough network capacity of 255 devices (including Access Point) on a single subnetwork.
One amendment under 802.11 heavily used in mobile automation applications is 802.11r, which is a provision for fast roaming between access points. 802.11 already has support for roaming but in normal operation, this requires re-authentication when transitioning between access points which can result in reconnecting times amounting to several seconds, which can be unacceptable in automation applications, particularly with control and safety data.
With 802.11r, the authentication key can be cached, reducing transition times to a few milliseconds (30-50ms typically). With the higher throughput wireless standards (802.11n and 802.11ac) plus 802.11r fast roaming, 802.11 has gained greater acceptance inside the factory.
802.11 Limitations and Considerations
Wireless is being effectively used today for challenging applications such as mobile automation, but like any communications technology, there are limitations in performance that must be considered when implementing such as the number of connected devices, signal interference, and physical obstructions. There are a couple of factors that affect the scalability of 802.11-based networks.
While the theoretical limit of a single IP-based subnetwork segment is 255 devices, the practical limitation is significantly fewer devices. This is due partially to the fact that 802.11 channels at 2.4 or 5 GHz are typically 20 MHz or 40 MHz wide and that all devices on the network need to share that channel. As the number of devices increases, smaller slices of bandwidth are available for each device to pass data through which can impact the performance of the overall system.
Specifically, for safety data, this may mean that longer timeouts need to be used and that the minimum safe reaction time may be longer than desired for the application in question. In those cases, the network may need to be de-scaled to allow for enough bandwidth to the devices on the network.
Additionally, 802.11 is based on publicly available, unlicensed frequencies, which means that the channels may be shared by multiple networks, further reducing the amount of available bandwidth. This is difficult to avoid at 2.4 GHz frequencies as there exist only four (4) non-overlapping channels that can be used for networks and it should be expected that 2.4 GHz channels will be shared.
Higher throughput channels in the 5 GHz spectrum are more readily available with fewer overlapping frequencies and are being increasingly used for critical applications. This will come under strain over time as more networks are configured in the 5 GHz space.
Reliability can also be a challenge when using 802.11-based networks. The biggest factor that impacts network reliability is signal interference, particularly at 2.4 GHz frequencies. As previously mentioned, these frequencies are available publicly. While 2.4 GHz is most closely associated with 802.11 networks, this frequency is used by other devices and communications technologies including cordless handsets, microwave ovens, Bluetooth, and Zigbee among others.
In addition to accounting for the density of networks in the spectrum, the noise generated by these other devices can have a significant impact on the quality of communications.
The presence of physical objects can also impact the reliability of wireless communications. In a perfect environment, wireless transceivers connect with each other over open air, but this is not practical in a manufacturing environment, as there will be machinery and building structures to contend with.
Absorption or reflection of wireless signals by structures can significantly reduce the effective range of these networks and create holes where signals may be lost completely. Materials such as solid metals and concrete provide the biggest challenges for wireless signals.
The practice of roaming between access points, which is to be expected in most manufacturing facilities, can also have negative impacts on critical communications. 802.11r mitigates this significantly, but if there are any issues during re-authentication that extend the hand-off significantly past the normal 30-50 ms timeframe, this can create packet loss and bit errors that can significantly impact control data and safety data.
Addressing Challenges with 5G Cellular
It is important to state that 5G cellular technology is still emerging and current implementations are limited, however, there is much that is built into this network that will address many of the challenges currently posed by other wireless networks. The three core technologies of eMBB (Enhanced Mobile Broadband), uRLLC (Ultra-Reliable Low Latency Communications), and mMTC (Massive Machine-Type Communications) will be critical for autonomous vehicles.
For throughput and reliability, eMBB delivers data rates around 10 Gbits/S, uRLLC delivers network latencies of less than 1ms and mMTC delivers high network density allowing for scalability. The first challenge to focus on is the complications posed by running critical communications for mobile robots in unlicensed frequencies used by 802.11.
With 5G, you have a mixture of licensed and unlicensed bands (5 GHz spectrum). One slice of spectrum to make note of in North America is the 3.5 (3.55-3.7) GHz band known as CBRS (Citizens Broadband Radio Service) which has traditionally been reserved for naval radar communications.
This has been released to be used for semi-private networks. Instead of having to request a license for spectrum from a government entity or purchase a private network from a public provider, manufacturers can request a band from a Spectrum Allocation Server that then assigns the band based on terrain and RF density calculations.
The benefit to the manufacturer is that if they are using lower-power radios, they can have a wireless spectrum that is specifically allocated to their facility without interference from other networks. This in combination with the high speed (<20 GBPS) and low latency (<1ms) characteristics of 5G makes this feasible for transmitting critical control and safety data.
Another challenge that 5G technology helps address is the impact of physical structures on reliable wireless communications. While any wireless signal is impacted by structures, 5G can use reflected signals in a beneficial way. 5G supports the use of a high-density antenna construct known as massive MIMO (Multi-Input Multi-Output) and beamforming to optimize connection pathways.
Transmitters can broadcast signals in multiple directions while the receiver sees multiple copies of the same signal which is then either destructively averaged if out of phase or constructively summed if in phase. This increases the reliability of these connections with the added benefit of increasing throughput as well. With more reliable connectivity, critical communications for control and safety can be maintained. Figure 1 shows a simplified view of a multi-path system of Base Stations (BS) and end devices (ex. AGV) in a manufacturing environment.
The last challenge we will address is handover times associated with roaming. While 802.11-based networks supporting 802.11r minimize handover times to a few milliseconds this can still impact critical communications. 5G technology addresses this by allowing devices to make connections to other base stations before separating from a previous base station as shown in Figure 2. This allows for seamless communications on the 5G network and eliminates the possible issue of timeouts and bit errors during handovers.
The final question is whether this can be scaled to meet a growing market for mobile robotics. While it is not expected that autonomous vehicles will be deployed to this scale, with support for mMTC (Massive Machine Type Communications) service on 5G, network density can achieve up to 10⁶ devices/km².
Conclusion
Wireless technologies are gaining acceptance in manufacturing as the need to increase capacity within a limited physical footprint drives the need for mobile robotic systems that can be flexibly deployed. With existing wireless technology still presenting some technical hurdles, maximizing the productivity of these systems can be challenging. 5G communications address many of the key challenges including scalability, latency, and reliability. This will have a significant long-term impact on communications in manufacturing. Download our expanded whitepaper to learn more:
More about connectivity for the automated warehouse