Delivery Robots

By integrating 3D vision sensors and Xinmai Robot Controllers, delivery robots can flexibly avoid obstacles and navigate complex restaurant environments, significantly improving meal delivery efficiency, enhancing the user dining experience, and substantially reducing operating costs for businesses.

Difficulty in delivering during peak hours has long been an industry pain point for delivery robots. Restaurants experience high foot traffic during peak hours, and the environment is highly complex and unpredictable.

Equipped with RGBD depth sensors and efficiently integrating RGB cameras and LiDAR, they possess powerful visual perception capabilities, allowing them to navigate freely even in highly dynamic and complex environments.

20220325-普渡巧乐送-4K-L.mp4_20221111_110957.445.png

20220325-普渡巧乐送-4K-L.mp4_20221111_110903.988.png

Delivery Robots

business.mp4_20221111_165600.026.png

Delivering meals to designated spots

What features are essential for an RK3588-based intelligent robot controller?

As the core brain of a robot, an intelligent robot control board needs to possess a series of key features to support its efficient and intelligent operation. The following are the main characteristics required for an intelligent robot control board:

NPU Acceleration

Supports offline large models: Capable of handling complex AI tasks and running large neural network models even in unstable network or offline conditions.
Object recognition and other algorithms: Built-in or loadable algorithms for object recognition, facial recognition, scene understanding, and more, enhancing the robot's perception capabilities.
Monocular ranging algorithms: Supports distance measurement using a single monocular camera, enhancing the robot's spatial awareness.
Flexible computing power stacking: Provides flexible computing power expansion capabilities, allowing NPU resources to be adjusted according to task requirements.

Operating System Support

Supports Linux and Android systems: Offers broad support for both Linux and Android operating systems, making it convenient for developers to choose the appropriate OS for development.
Android-friendly UI design: Provides user-friendly interface design tools and frameworks for robots utilizing the Android system.
Linux standard adaptation for ROS/ROS2: Seamlessly compatible with ROS (Robot Operating System) and ROS2, providing robust ecosystem support for robot application development.

Intelligent Voice Interaction

Multi-microphone array: Configured with multiple microphone arrays to improve voice capture clarity and directionality.
Offline voice wake-up word: Supports offline voice wake-up functionality, allowing the robot to be activated by a specific wake-up word even without an internet connection.

AMP Architecture Support

Strong real-time performance: Ensures the robot control board has extremely high response speed and real-time capabilities when processing critical tasks.
High stability: Employs a stable hardware and software architecture design to reduce system failure rates and enhance the robot's operational reliability.
Lower hardware cost: Achieves a balance between performance and cost by optimizing the AMP (Asymmetric Multiprocessing) architecture.

ROS Compatibility

Compatible with standard ROS/ROS2: Seamlessly integrates with ROS and ROS2, allowing robots to easily access various libraries, tools, and applications within the robot ecosystem.
Compatible with various radars: Supports multiple radar sensors (e.g., LiDAR, millimeter-wave radar), enhancing the robot's perception capabilities in complex environments.
Compatible with RK structured light: Highly compatible with Rockchip's self-developed structured light modules, enabling more precise 3D perception and measurement.

Advantages of RK3588 in Robot Control Boards

High CPU computing power: Up to 93KDMIPS of CPU computing power, providing the robot control board with robust data processing capabilities, ensuring the robot can quickly respond to various commands and tasks.
High GPU computing power: Up to 512GFLOPs of GPU computing power, supporting complex graphics rendering and multimedia processing, enhancing the robot's visual perception and interactive experience.
High NPU computing power: Triple-core NPU with up to 6TOPs (INT8) of computing power, supporting deep learning frameworks and providing efficient and precise computing support for AI applications on the robot control board. For instance, with the YOLO-v8n model, a single core can achieve an inference speed of 59.6fps, meeting the requirements of real-time applications.
Rockchip self-developed modules: Rockchip's self-developed structured light modules and RK1808 computing stick, among other modules, provide more specialized hardware support for the robot control board, enhancing the robot's overall performance and functionality.
Industrial-grade chip standard: Supports a wide operating temperature range from -40°C to 85°C, ensuring stable operation of robots in various harsh environments.

RK3588-based Robot Control Board Design Block Diagram