SOLUTION ARCHITECTURE
AIB-MX23: The Core Brain for Autonomous VGV Intelligence
Mounted inside the VGV chassis, the Aetina’s DeviceEdge AIB-MX23 Edge AI platform is the robot's onboard brain, powered by NVIDIA Jetson AGX Orin 64GB — its 2048-core Ampere GPU delivers up to 275 TOPS of AI performance, while 64 GB of 256-bit LPDDR5 memory at 204.8 GB/s keeps sensor fusion, SLAM, and navigation running concurrently without contention.
At the foundation is multimodal sensor fusion. The AIB-MX23 integrates 2D LiDAR to define real-time obstacle safety zones, 3D LiDAR to feed a SLAM pipeline for centimeter-level localization, and FRAMOS D455e cameras for visual odometry and depth perception. Wheel encoder feedback anchors the fusion loop, correcting positional drift between scan updates. These fused outputs feed navigation algorithms accelerated by NVIDIA® CUDA® and TensorRT™ on the Jetson AGX Orin's module — computing optimal routes, generating avoidance trajectories, and issuing speed and steering commands within the tight latency that safe warehouse operation demands. A dedicated stability loop runs in parallel, modulating speed and turn radius to keep the trailing cart train moving smoothly at scale. Each AIB-MX23 also has Wi-Fi and 10GbE LAN links to the ground control system, enabling coordinated fleet operation across Anantak's over 200-unit deployments.
Compact, rugged, and rated for −25°C to +80°C operation, the AIB-MX23 is purpose-built for the demands of continuous industrial deployment.
Sensor Inputs