Centralized Robotic Navigation System: The Happy Medium Between QR Code Navigation and SLAM?

In August 2022, 634AI announced it had been contracted to deploy 200 Autonomous Mobile Robots (AMRs) across Musashi Seimitsu’s 35 manufacturing facilities worldwide. Based in Israel, 634AI focuses on building indoor navigation and Real-Time Location Service (RTLS) technology for industrial and manufacturing purposes.

The partnership with Musashi Seimitsu is focused on MAESTRO, a proprietary AI-powered control tower system developed by 634AI to manage the movement of autonomous robots in an indoor environment. MAESTRO offers a constant visual mapping of the entire floor, with real-time location of critical assets, such as Autonomous Guided Vehicles (AGVs), AMRs, and forklifts. This feature enables end users to track raw material movements, provide productivity and utilization data, recognize and prevent hazards and obstacles, provide safety alerts for forklift drivers, and even navigate the movement of heterogeneous autonomous robot fleets.

Currently, most indoor AMRs are deployed using two approaches: infrastructure heavy and infrastructure free. The first approach (infrastructure heavy), predominantly adopted by AGVs, is QR code navigation. End users need to deploy QR codes or other fiducial markers across the operational site floor. AGVs will scan the markers to help them localize and identify the direction to their destination. Despite being relatively cost-efficient compared with other approaches, this navigation method requires greenfield deployment in a highly structured site. Therefore, it cannot work in more structurally challenging locations, like a mezzanine, and it cannot collaborate with human employees.

In comparison, robots that utilize either vision-/camera-based SLAM or LiDAR-based SLAM (predominantly AMRs) are ideally designed to work in a highly unstructured environment in collaboration with human employees. Considered the infrastructure-free approach, robots in this environment are equipped with RGB and stereo cameras, proximity sensors, and 2D and 3D LiDAR onboard to identify and analyze their surroundings, enabling them to perform mapping, localization, path planning, obstacle avoidance, and object detection. Not surprisingly, all these computational workloads take place on robots, leading to high computational requirements. Therefore, these AMRs operate as stand-alone robots with sophisticated intelligence supported by high-performance processors from Intel, NVIDIA, Qualcomm, and Xilinx.

Configuring AMRs to work in tandem with one another can be a complex task for integrators, especially if AMRs from different vendors are being used within the same facility. In some cases, application programming interfaces (APIs) have enabled various AMRs to communicate with one another, but this type of API remains incredibly complex and challenging.

634AI comes in with an infrastructure-lite approach—a medium between the two methods mentioned above. Essentially, 634AI deploys a camera system onto the ceiling of the factory. The cameras capture and stream video into a central system that then acts as a control system by stitching all the videos together, guiding the robots. This system possesses a bird-eye view of the entire operation, enabling real-time floor mapping for employees, materials, and equipment.

As ABI Research foresees global annual shipments of manufacturing and warehousing mobile robots going from 212,000 in 2021 to over 1.3 million in 2030, the infrastructure-lite approach from 634AI brings unique advantages. Robots can coexist with human employees without requiring expensive onboard hardware. At the same time, instead of investing in a myriad of robotics systems from different vendors, end users can deploy robots from multiple vendors and have them centrally managed by a single system. RTLS capabilities also ensure worker and equipment safety.

At the moment, only a handful of vendors can provide an integrated multirobotics system. 634AI’s approach would free end users from potential vendor lock-in for those looking to integrate best-of-breed robotics automation for different functions.