We will be reviewing a highly efficient edge AI solution for autonomous commercial mowing. Lanner the hardware LEC-2290, an edge AI computing power and MoV AI, provides the software, a robotics engine platform, designed to empower AMRs (Autonomous Mobile Robots).
Autonomous Mowing Requirements.
Creating a safe and productive Autonomous Mobile Robot (AMR) to mow the lay is, in no particular way, an easy task. Two of the most basic commercial autonomous mowing systems requirements are safety and productivity. Below are further details for these requirements.
Having a robot running with blades and cutting grass around home yards, sports fields, or parks where people and pets frequent makes safety paramount. An AMR-based autonomous lawn mower should have various sensors (and possibly cameras) to navigate more efficiently, controlling its relative speed and distance to any object blocking the path. Optimally, AMRs must be capable of identifying objects, making quick vital decisions, and steering clear or immediately stopping.
Safety entails that the autonomous mowing system must be intelligent enough to process large amounts of real-time sensor or image data. When the system’s brain is not embedded in the robot but in a nearby server, the wireless communications must be reliable and fast. The robot must also have remote control capabilities for added redundancy and safety.
An autonomous mowing mobile robot needs to be constantly aware of its location, speed, and other variables to cut the grass productively. The mowing robot needs to control different parameters like 4-Wheel-Steering, 4-Wheel-Drive, and its deck size to mow in challenging areas. Plus, allowing a human to program and control it remotely can make the task more productive.
Autonomous mowing robots require fast and optimal decision-making to allow them to control all these different parameters. Although the intelligence and control could be delivered from remote servers like the cloud, in reality, mowing machines do not always have access to the best bandwidth and lowest latency networks for accessing the cloud. The brain of the mowing system must be brought on site.
Solution: Edge AI for Autonomous Mobile Robot (AMR)
Lanner’s LEC-2290 hardware and MoV AI, a robotics engine platform, partnered to create a bundle solution. The bundle solution provides highly efficient edge AI capabilities for any commercial AMR, including autonomous mowing. With edge AI, all AI computation is performed at the network edge (near the user, data, or application) rather than in the cloud or a private data center.
Lanner’s LEC-2290 is an intelligent edge AI appliance powered by the Intel® Core™ i7-8700T/i7-8700 CPU (codenamed Coffee Lake S) and two DDR4 with a max. 32GB of RAM. The appliance is designed for AI video analytics or factory machine vision. It is Nvidia NGC-ready and can be expanded with GPU, VPU, or FPGA (via two PCIe). It also comes with two modules to expand the appliance to LTE and 5G wireless support.
The edge AI appliance also features abundant I/O ports, including two RJ45 GbE LAN, four PoE, four USB3.0, six COM ports, eight DI & eight DO, and two removable HDD/SSD external slots with RAID.
The LEC-2290 has partnered with MoV.AI in software for edge AI applications. MoV.AI is a (Robot Operating System) ROS-based robotics engine platform designed to help AMR manufacturers and integrators to build, deploy and operate robots. The MoV platform includes a ROS-based IDE, autonomous navigation algorithms, deployment, operational tools, cybersecurity, flexible interfaces, and APIs.
How does it work?
AMRs combine hardware, software, and distinctive off-the-shelf (use-case) components. The autonomous mowing robot leverages LEC-2290 for edge AI (hardware), the robotics engine platform— MoV.AI (software), and a series of sensors (components), including LiDAR, GNSS, ultrasonic, or cameras. The autonomous mowing robot can use a pre-defined map and path with satellite data and its navigation system to move safely and productively. The mowing robot will cut the grass within a defined boundary and across a specified path.
As shown in the picture below, the commercial autonomous mowing system can also be connected to a remote operation and reporting center via 4G or 5G.
AMRs can also leverage imaging data collected from their mounted video cameras. These cameras can send collected images to a computer vision system to get more precise navigation. The IP cameras can send data nearby to an AI video analytics appliance such as LEC-2290.
AMRs provide advanced robotics for an underserved market: commercial mowing. They can move around limited areas like yards, sports areas, parks, etc., independently.
On a basic level, AMRs can mow the grass safely by knowing their navigation map and its boundaries. But most significantly by having sensors, cameras, or a machine vision to detect objects (people, pets, shrubs, balls, etc.) and make vital decisions on the go. Edge AI-empowered AMRs can also be more productive, not only by reducing labor workloads but also by introducing optimizations. In addition, AMRs can improve schedules, alternate mowing paths, create reports, and more— all with the right software and intelligence.
- Cost efficiency. AI/ML, Deep Learning, and Computer Vision have always been expensive solutions. But now, by bringing all this intelligence closer to where data is (rather than bringing data to where AI is, usually hosted in the cloud), costs can be optimized. There can be a significant reduction in network and storage demands.
- Introduction to AI. With AI at the edge, AMRs can learn from their experience, adapt their behavior, and easily adjust to new inputs. For instance, the AMR can only operate when workers are off, when there is no rain, learn to adapt its deck, etc.
- Introduce side benefits: Having an autonomous moving system can introduce new services and applications, such as remote surveillance, water level controls, anti-theft sensors, etc. One of the most significant benefits is introducing computer vision (CV)— giving the mowing robot a new pair of eyes!
For more information on the Edge AI computer or the highly efficient edge AI for commercial autonomous mowing solution, please contact Lanner’s sales representative.
Intelligent Edge Computing Box PC w/ Support for Intel® Core™ i7-8700T/i7-8700
|Support Intel® Core™ i7-8700T/i7-8700 Core i (FCLGA1152), Codenamed Coffee Lake S