Atlas Mba'yru Mopu'ãha Iñarandúva (Forklift Inteligénte)

Press Kit

Unique Properties

Atlas is an industrial grade smart forklift embodied robot designed for warehouse and manufacturing environments. Integrating BEV perception, AI Agent planning, and end to end execution capabilities, it forms a complete closed loop from environmental perception to task comprehension to autonomous operation, optimizing goods handling processes and addressing the complex deployment and rigid usage typical of traditional autonomous mobile robots.

Dimensions & Packaging

1206 mm x 1542 mm x 1727 mm

Tags

Smart Forklift, BEV Perception, AI Agent Autonomous Planning, Flexible Logistics, Natural Interaction, End to End Execution, Modular Deployment, Warehouse Automation

Production Technology

The Atlas chassis combines a high strength aluminum alloy frame with composite protective panels for optimal weight durability balance. Its cameras and LiDAR are housed in a roof mounted perception tower with active dust removal and temperature control, ensuring industrial grade reliability. Wireless charging allows uninterrupted power replenishment during operation, boosting continuous efficiency.

Design Challenge

Handling uncertainty is considered as the biggest challenge. Industrial sites are rife with unpredictability, goods slightly displaced, temporary water on the floor, personnel unexpectedly entering the area. Traditional automation relies on strictly regulated environments, whereas Atlas must actively adapt to complex changes. A probabilistic decision making model and continuous learning mechanism enable Atlas to grow smarter with each unexpected event.

Research

Atlas features the integrated EmboSense sensor, unifying a 3D LiDAR with four fisheye cameras. This delivers 360°environmental perception free of blind spots. Coupled with vision algorithms, it produces high precision panoramic view and colored Gaussian spheres, enabling precise identification of shelves, goods, aisles, and other elements. The system also supports end to end object recognition and continuous self learning iteration.

Inspiration

Existing industrial autonomous mobile robots often rely on preset environments and fixed workflows, making them not suited for dynamic, complex real world factory operations. Atlas is designed to fundamentally transform the inflexibility of conventional automation solutions. It can adapt flexibly to non standardized tasks much like an experienced worker, thereby enhancing both the adaptability and efficiency of logistics automation.

Project Duration

The project commenced in Shanghai in March 2024, was completed in Shanghai in May 2025, and was exhibited at CeMAT Shanghai 2025.

Operation Flow

Users can issue task commands to Atlas via natural language or a graphical interface. The robot autonomously interprets task intent, plans paths, and executes operations. Throughout, intuitive UI, voice feedback, and status indicators facilitate human robot interaction, enabling deployment and application with zero technical barriers, no specialized programming required.

Image Credits

Fan Wu