Six surround-view cameras and BEV+OCC algorithms have been added, integrating LiDAR data to provide full vehicle coverage and eliminate blind spots.

The position of the hook is detected and identified using tags on the visual sensor mounting ring; subsequently, through path planning and targeting, the vehicle is guided to move along the planned path to automatically complete the hooking operation.

Identify the location of available charging stations using visual sensor detection or positioning solutions to enable automatic vehicle charging.

When the vehicle encounters a malfunction and reports the fault information, the monitoring center can conduct remote 3D surround-view video monitoring of the unmanned logistics robot.

OTA updates and out-of-the-box readiness.

Replace the H16 with a solid-state LiDAR (such as the E1R); replace units like the INS570D.

The perception module implements dual-modal fusion perception using a forward-facing E1 solid-state LiDAR and a front-view camera, significantly improving 3D object detection metrics.

Local Planning — Stage 1: Perform rule-based obstacle avoidance and maneuvering around obstacles when conditions permit; Stage 2: Navigate through efficiently.

Following the architectural adjustment, it is adaptable to various platforms, and resource consumption is significantly reduced.

Port the Tianzhun Orin solution to other platforms (such as the RK3588) as a cost-effective alternative.

Deploy large-scale decision-making models to make vehicles more intelligent, enabling them to flexibly handle various unknown scenarios.

Built using large language models, the system supports the issuance of instructions and tasks via natural human language.

Enables precise identification and fork-handling of non-standard load carriers. Utilizing 3D sensors, the system performs cargo, pallet slot, and target area detection, while supporting control functions such as automatic pallet slot alignment and dynamic alignment with the target area.

Supports OTA updates.

Based on multi-sensor fusion perception combining vision and LiDAR, the system is capable of detecting, tracking, and predicting the trajectories of obstacles along the planned route, as well as identifying traversable areas in the vicinity; it supports navigation, obstacle avoidance, and obstacle bypassing along the planned route.

Unmanned forklifts can perform mapping and route data collection in remote-control driving mode.

Supports high-precision positioning with an accuracy of within 2 cm.

The decision-making module can execute standard task planning and also supports an architecture that integrates Large Language Models for task planning.