AMR - what does it take to control it?

AMR - what does it take to control it?

AMR - what does it take to control it?

Autonomous Mobile Robots (AMR) navigate autonomously and intelligently through warehouses and the shop floor, check vegetables, fruit and berries for pests in the greenhouse and transport online packages the last mile to the recipient. The fleet management software guides each robot to its destination, assigns tasks, checks battery status, calls the AMR to the charging station and monitors routes and order status in real time. Comp-Mall has a control PC for this in its range.

On the one hand, AMRs have a lot of technology on board; on the other hand, they are battery-operated and have to call the charging station regularly. The control PC, usually a GPU computer, must therefore represent the best possible compromise between performance and energy consumption.

Tested in practice

Already in use with autonomous mobile robots is the GM-1000, a GPU computer with FCLGA1151 socket for 8th generation Intel processors. The Core i7-8700 CPU with six cores and 4.6 GHz maximum Turbo clock frequency generates a power loss of 65 watts, the energy-optimised Core i7-8700T CPU with six cores and 4.0 GHz maximum Turbo clock frequency only 35 watts. From a total of twelve CPU variants, the suitable solution can be found for the optimal balance between performance and energy consumption.

Computing and graphics power can also be outsourced to a dedicated MXM graphics card, which is available in several variants. For example, the MXM-A2000 from NVIDIA with 2560 CUDA cores and a peak performance of 8.25 TFLOPS in single precision floating point calculations. In addition, 80 tensor cores and 20 ray tracing cores are available. The MXM-A2000 graphics card supports the DirectX 12 and OpenGL 4.6 graphics APIs and the CUDA Compute 8.0 and above and OpenCL 1.2 computation APIs. The GDDR6 memory with a bandwidth of 192 gigabits per second has 8 GB VRAM.

To navigate, AMR must recognise the environment. For this purpose, AMR use various techniques such as LiDAR, laser scanners, 3D cameras. During the journey, the AMR performs scan matches to determine the exact location. The object being searched for must also be identified - most AMR use laser scanners for this. The data is delivered to the computer via CAN bus, processed and transmitted to the central fleet manager, which integrates the autonomous mobile robots into the environment, coordinates the transport orders, monitors the battery status, plans the optimal route and much more. The "connection" between the fleet manager and the individual AMR takes place either via WLAN or mobile radio. The GM-1000 offers both options.

All info at: