The Mobile Robot Space is Quickly Getting Crowded: Collaboration, Commoditization, and Consolidation to Follow

Based on a recent ABI Research report (AN-4946), Autonomy Solution Providers (ASPs) are beginning to compete extensively in the mobile robot space, adding vendor-independent technology stacks for navigation, perception, motion control, and general mobility to the mobile robotics market. Facing them are Mobile Robot Providers (MRPs) like Fetch Robotics, Waypoint, and more traditional vendors who are focusing on integrated robotic platforms as opposed to retrofitting forklifts, Tuggers, industrial floor cleaners, et al.

As the state of the robotics industry has progressed, the availability of autonomous software for mobile robotic platforms has increased. During this time, the primary benchmark for autonomy on ROS robots has been the standard GMapping package, documented on the ROS Wiki, which is a wrapper for GMapping from OpenSLAM. GMapping has one of the most widely forked ROS Perception packages in existence, and is the default go-to method for starting basic autonomy and SLAM work.

While the GMapping method has been widely used for academic papers (with over one thousand references to it via Google Scholar), it has widely been acknowledged that it is not the most accurate method of localization for autonomy.

In 2017, Clearpath Robotics packaged this software into the Autonomy Research Kit (ARK), a black-box secured solution that includes a dedicated SLAM computer and forward and rear facing LiDAR. Currently, ARK is only available on Clearpath Robotics’ commercial line of robotic platforms, but it has developed further proprietary offerings for GNSS teleoperation.

Seegrid and Brain Corp have developed further at this stage. Seegrid was one of the first commercial developers of vSLAM, which uses stereoscopic camera arrays and 3D SLAM as opposed to Lidar. This serves as a platform for advanced machine vision, infrastructure-free navigation, and on-device autonomy. Seegrid has since become the dominant player in the high-payload, large form factor (LFF) material handling market. On the other hand, Brain Corp is dominating industrial floor cleaning and pushing into new verticals like consumer, hospitality, and security. Both provide autonomy stacks with varying types of hardware. These proprietary systems have lower rates of error than GMapping and are more intuitive to the user. As prepackaged solutions, they are better placed to be deployed at scale in busy working environments.

As mobile robots multiply, the emphasis on the single robot will be replaced by the idea of multiple robots as systems. As these robots increasingly work as groups and swarms, emphasis will move away from advanced onboard cameras, arms, and individual capability and toward connectivity and localization for mass-autonomy. This requires infrastructure that provides cm to mm level localization for robots. Ultra-Wideband (UWB) technology, which has long been deployed to provide high location accuracy for mines and train systems, is best placed to fulfill this purpose. The low-frequency of UWB makes it preferred to Wi-Fi or Bluetooth in metal-rich and high-disturbance environments, as it is less susceptible to disruption.

Among the key enablers of this technology is Quantitec, a German company with a fully integrated platform called IntraNav. IntraNav is compatible with the default open source robotics platform Robotic Operating System (ROS). This interoperability allows robotics manufacturers and end-users utilizing ROS to integrate cm-level accuracy through location data with the ROS Simultaneous Localization and Mapping (SLAM) software, allowing for increased technical redundancy of robotics movement and the ability of robots to navigate their environment safely and in tight spaces with multiple platforms.

From 2015, it was fairly clear that iRobot was developing an advanced localization solution, as it was in talks with the Federal Communication Commission (FCC). In early 2019, it was confirmed that iRobot had incorporated UWB beacons into their latest robotic lawnmower solution, replacing the more expensive and cumbersome wired geofencing solution that primary robotic lawnmower manufacturer Husqvarna uses.

UWB technology will initially be used for cm-level localization but is gradually being developed to a level of sophistication at which it will be able to create 3D digital representations of robots and infrastructure in real-time, with mm-accuracy opening the possibility of external beacons as an alternative to embedded machine vision, which is far more compute-intensive. MRPs and ASPs that link up with the key UWB providers will have a strong platform to then develop advanced value-added services, including digital twins, superior fleet management systems, and complex tasks for robots like mobile manipulation and vehicle-to-vehicle collaboration.

The key challenge for UWB technology continues to be its price. The sole major supplier for UWB chipsets is Decawave, and the lack of effective competition is one of the key things preventing significant price-decrease. Current chipsets for UWB cost around US$4-5, as opposed to equivalent components for RFID or BLE beacons that are valued in cents.

Years ago, Amazon was confident about the possibility of drone delivery. ABI Research is skeptical about the viability of this opportunity, given myriad regulatory and technological hurdles, particularly relating to infrastructure, safety, and privacy issues, and is unsurprised that Amazon quietly acquired ground-based last-mile delivery robot maker Dispatch as an alternative. Last-mile delivery for ground mobile robots will have many of the same challenges as drone delivery, but ASPs in this space are developing the sophisticated SLAM capabilities that make them great technology providers to automotive companies and self-driving car startups. For example, last-mile delivery robot maker Nuro has provided navigation software to self-driving truck startup Ike.

Generally, Amazon’s partnerships and acquisitions with robot companies have preceded a big play in offering new robotics. The addition of AWS RoboMaker will provide greater architecture for this ecosystem. Given that Amazon recently partnered with and promised to buy products from French ASP Balyo, it is expected it will go on to acquire the company, or develop their own AMR offering for the industrial and commercial space, to complement their strength in AGV offerings.

As Amazon’s portfolio grows, the rest of the market will consolidate. Companies like Seegrid, Brain Corp, Vecna and Fetch are developing significant partnerships and deploying on increasingly impressive scales. As they continue to place more emphasis on software and value-added services, the ability of other, smaller MRPs and budding ASPs will be tightened. As value shifts from the robots themselves to AI, back-end infrastructure, and the services they provide, 2019 will be the year that the mobile robotics explosion finally consolidates.

Collaboration and integration between and with these frontrunners is therefore key. ASPs and MRPs should cooperate on developing more fool proof and dynamic SLAM solutions; those that can seamlessly provide indoor and outdoor autonomy, for example, will have a huge advantage. Canvas Technology, PerceptIn, Nuro, and Autonomous Solutions are the front runners in providing outdoor navigation at the non-automotive level and will be key to future innovation.

The value of UWB has been mentioned, but there are also opportunities in wireless charging solutions and cloud systems for robots. Those companies that broaden their portfolios and successfully make the ecosystem play are best positioned to compete with Amazon, the most ferocious competitor in a number of markets for the last 5-10 years.