Hiwonder ArmPi Ultra ROS2 Robot Arm with Raspberry Pi5 Options

OverviewThe Hiwonder ArmPi Ultra is a programmable desktop robot arm for ROS2 education, vision-guided tracking, grasping and sorting. Its six smart bus servos and metal structure support manipulation experiments, while Raspberry Pi5 and STM32 divide higher-level processing and servo control. Choose the kit-specific camera and voice hardware, then select a version with or without Raspberry Pi5.3D perception requires the depth-camera configuration.Features All-metal aluminum-alloy structure with anodized finish and bearing-supported base. Six intelligent bus servos with separate Raspberry Pi and STM32 control roles. App,PC software,Python and ROS topic/node control. Inverse kinematics,DH modeling,simulation and programmable vision tasks. Kit-dependent monocular/depth vision and Advanced voice-interaction hardware. Software and Manipulation Functionscomprehensive tutorialsdesigned to inspire and support your AI-driven creative projects.Depth Vision, 3D Scene Flexible GrabbingDepth-camera configurations support target recognition and tracking, RGB+D fusion detection, and object grasping in3D scenes.All-metal Structure, Bearing BaseThe body of the robot arm adopts an all-metal structure, and the anodized surface makes it exquisite and beautiful. The base uses industrial-grade bearings to meet high-demand grabbing projects.Intelligent Bus ServoArmPi Ultra robotic arm body is equipped with 6 high-torque smart bus servos, allowing the robotic arm to grab heavy objects and run smoothly.WonderEcho Pro AI Voice BoxThe Advanced Kit includes a WonderEcho Pro AI voice box that delivers excellent noise reduction and clear audio capture. It supports advanced features such as speech recognition, voice broadcast, and voice control.Dual-Controller Design for Efficient CollaborationArmPi Ultra combines the power of Raspberry Pi 5 with an STM32 controller to deliver smarter, faster, and smoother performance. Raspberry Pi 5 takes care of vision and decision-making, while STM32 ensures precise servo control and data process. Working together, they make the robot more responsive and capable than ever.Function ListLarge AI Model–Driven Embodied AI ApplicationsArmPi Ultra integrates multimodal AI model for cross-modal perception, allowing seamless interaction across text, voice, and image, delivering richer and more creative user experiences.Voice ControlPowered by ChatGPT, ArmPi Ultra can understand spoken commands and perform corresponding actions, delivering a natural and intuitive voice interaction experience.Color SortingWith the integration of ChatGPT and AI vision, ArmPi Ultra not only responds to voice commands but also accurately detects object colors within its field of view and sorts them automatically.Waste SortingBuilt with the YOLO deep learning model, ArmPi Ultra can recognize different types of objects in its environment. With ChatGPT integration, it interprets user commands and executes tasks such as intelligent waste sorting.3D Object SortingEquipped with a 3D depth camera, ArmPi Ultra detects objects of various shapes within its field of view. Combined with ChatGPT, it understands user instructions and performs the required sorting actions.AI Embodied Intelligence ApplicationsPowered by high-performance hardware and multimodal large model, ArmPi Ultra can perceive and understand the environment, plan actions, and perform tasks with flexibility. It enables advanced human–robot collaboration and intelligent task handling, delivering a more natural and efficient interactive experience.Real-Time DetectionPowered by ChatGPT, ArmPi Ultra understands user commands and analyzes scenes in real time. It provides instant feedback on image content and object features through both text and voice.Vision TrackingBy combining ChatGPT with OpenRouter's visual large model, ArmPi Ultra interprets user commands to identify and locate objects. With inverse kinematics algorithms, it achieves precise and responsive target tracking.Intelligent TransportWith the support of the Vision AI large model, ArmPi Ultra identifies target objects and retrieves their positional data. Combined with ChatGPT, it interprets user commands to transport objects to designated locations.Smart Home AssistantPowered by a multimodal large model, ArmPi Ultra recognizes and analyzes objects in its environment. Paired with ChatGPT, it understands user instructions and provides appropriate responses through both speech and actions.Deep Learning Model TrainingCustom object-detection models can be trained and deployed using suitable supported hardware and software. Hardware-specific acceleration requires a compatible deployment environment; no onboard NVIDIA GPU is included.Waste SortingArmPi Ultra robotic arm kit is equipped with garbage pattern blocks. By loading the corresponding model, ArmPi Ultra can quickly recognize different garbage and place it in the corresponding classification area.Item SortingBy training models of daily items and generating corresponding models, with the support of depth cameras, ArmPi Ultra can quickly recognize and grab corresponding items by obtaining the depth information of the items.Gazebo SimulationThe ArmPi Ultra robotic arm is developed using the ROS framework and supports GAZEBO simulation. The robotic arm is controlled and algorithm verified in a virtual environment, which reduces the requirements for the experimental environment and improves experimental efficiency.ApplicationsDevelop color-sorting exercises,vision tracking,depth-guided pick-and-place and simulated manipulation experiments. Use depth-equipped kits for3D spatial tasks and Advanced hardware for voice-box activities. Network AI services require compatible accounts,connectivity and software configuration.What's IncludedOne selected ArmPi Ultra kit and its kit-specific accessories. Starter uses a monocular camera; Standard includes depth vision; Advanced adds the WonderEcho Pro voice box. Raspberry Pi5 is included only when explicitly selected. Refer to the corresponding packing panel for the controller,leads,mounting hardware and learning accessories; demonstration computers are not included.FAQDoes every kit include a depth camera?No. Starter uses a monocular camera; Standard and Advanced use depth cameras.Is Raspberry Pi5 included?Only options explicitly naming Raspberry Pi5 include the selected memory configuration.Which kit includes the AI voice box?The Advanced Kit includes the WonderEcho Pro voice-interaction box.Configuration Guide Kit Vision and voice configuration Starter Kit USB monocular camera; no3D depth camera or WonderEcho Pro voice box. Standard Kit Depth camera for3D perception; no WonderEcho Pro voice box. Advanced Kit Depth camera and WonderEcho Pro AI voice interaction box. The Raspberry Pi5 selection is independent of the kit: choose without Raspberry Pi5,or with2GB,4GB,8GB or16GB memory. A version without Raspberry Pi requires a compatible controller and software setup supplied separately.3D vision functions require the depth camera;voice-box functions require the Advanced Kit hardware and appropriate software/services.Technical Specifications and Configuration References Aurora930 Pro Specifications Module parameter Size 76.5 × 20.7 × 21.8 mm Imaging performance Depth data format 16-bit Raw Baseline 40mm Depth resolution/Frame rate 640×400 @12fps (FOV: 74°×51°) Interface USB 2.0 Wafer connector RGB data format NV12 Depth accuracy ±8mm @1m RGB resolution /Frame rate 640×400 @12fps (FOV: 74°×51°) Working distance 15–300 cm IR data format 8-bit Raw Operating temperature -10℃ to 55℃ IR resolution /Frame rate 640×400 @12fps (FOV: 74°×51°) Operating humidity 0% to 95% RH (non-condensing) Firmware capabilities Firmware upgrade Supports USB OTA Update Operating illuminance 3–80,000 Lux Hot start delay <300ms Power supply 5V±10%, 1.5A System compatibility OS compatibility Linux / ARMv8 / ROS / Windows Power consumption Average <1.6W Safety rating Class 1 Laser safety Aurora930 Pro Specifications Module parameter Size 76.5 × 20.7 × 21.8 mm Baseline 40mm Interface USB 2.0 Wafer connector Depth accuracy ±8mm @1m Working distance 15–300 cm Operating temperature -10℃ to 55℃ Operating humidity 0% to 95% RH (non-condensing) Operating illuminance 3–80,000 Lux Power supply 5V±10%, 1.5A Power consumption Average <1.6W Safety rating Class 1 Laser safety Imaging performance Depth data format 16-bit Raw Depth resolution/Frame rate 640×400 @12fps (FOV: 74°×51°) RGB data format NV12 RGB resolution /Frame rate 640×400 @12fps (FOV: 74°×51°) IR data format 8-bit Raw IR resolution /Frame rate 640×400 @12fps (FOV: 74°×51°) Firmware capabilities Firmware upgrade Supports USB OTA Update Hot start delay <300ms System compatibility OS compatibility Linux / ARMv8 / ROS / Windows Product Parameters Product ArmPi Ultra Size 288*177*450mm Weight 1.47kg (Standard kit) Controller STM32F103RBT6 Material Full metal aluminum alloy bracket, sandblasted and anodized finish Control methods ROS topic/node control, Python program control, app control, PC software control Power supply 7.4V 6A DC Storage 64GB SD Card Software PC software + iOS / Android app Communication USB 2.0×2; USB3.0×2; serial port Camera USB Monocular Camera / Aurora Depth Camera OS RaspberryOS + Ubuntu 22.04 LTS (Docker) Programming tools Python /C/C++ /Javascript Servo models HTS-16L, HTS-25L, HX-06L, LX-15D, LX-225 intelligent bus servos Tutorials User manual, comprehensive tutorial, ROS source code, STM32 source code, system images, software Packaging weight & dimensions 3.2kg, 50x30x16cm; Item Specification Product ArmPi Ultra Size 288*177*450mm Weight 1.47kg (Standard kit) Controller STM32F103RBT6 Material Full metal aluminum alloy bracket, sandblasted and anodized finish Control methods ROS topic/node control, Python program control, app control, PC software control Power supply 7.4V 6A DC Storage 64GB SD Card Software PC software + iOS / Android app Communication USB 2.0×2; USB3.0×2; serial port Camera USB Monocular Camera / Aurora Depth Camera OS RaspberryOS + Ubuntu 22.04 LTS (Docker) Programming tools Python /C/C++ /Javascript Servo models HTS-16L, HTS-25L, HX-06L, LX-15D, LX-225 intelligent bus servos Tutorials User manual, comprehensive tutorial, ROS source code, STM32 source code, system images, software Packaging weight & dimensions 3.2kg, 50x30x16cm; Configuration GuideThe Raspberry Pi5 selection is independent of the kit: choose without Raspberry Pi5,or with2GB,4GB,8GB or16GB memory. A version without Raspberry Pi requires a compatible controller and software setup supplied separately.3D vision functions require the depth camera;voice-box functions require the Advanced Kit hardware and appropriate software/services.Technical Specifications and Configuration References Visual Details Features and Design ArmPi Ultra has 6 degrees of freedom. ArmPi Ultra is identified as a large AI model ROS robot arm. ArmPi Ultra is identified as a large AI model ROS robot arm. ArmPi Ultra supports 3D scene gripping and sorting. The host-controller architecture uses Raspberry Pi 5; board inclusion depends on the controller selection. The Advanced Kit supports multimodal AI model functionality through the Raspberry Pi 5 host. The host-controller architecture uses Raspberry Pi 5; board inclusion depends on the controller selection. The Advanced Kit supports multimodal AI model functionality through the Raspberry Pi 5 host. The Advanced Kit supports large AI model voice interaction through WonderEcho Pro. The 3D depth camera supports face tracking. The Advanced Kit supports large AI model applications. The Advanced Kit supports multimodal model applications. ArmPi Ultra integrates a large AI model for embodied AI applications. ArmPi Ultra's 3D depth camera includes an RGB lens. ArmPi Ultra uses OpenCV as its image-processing library and supports AI image recognition. ArmPi Ultra provides inverse-kinematics analysis, DH coordinate models and robotic-arm function interfaces. ArmPi Ultra uses an inverse-kinematics algorithm for coordinate-based movement in a 3D scene. ArmPi Ultra uses the MediaPipe development framework for gesture recognition, motion control, and face tracking. ArmPi Ultra recognizes various hand gestures for gesture-based control. The robotic arm model can be visualized in RViz with interactive position and orientation controls. ArmPi Ultra integrates MoveIt simulation for robotic arm control and algorithm validation in a virtual environment. ArmPi Ultra provides a scale-accurate URDF kinematics simulation model of the physical robot. ArmPi Ultra supports control through the WonderAi App. The WonderAi App includes a Position Calibration mode. ArmPi Ultra uses an Aurora930 Pro depth camera. The Aurora 930 Pro depth camera integrates a high-performance ISP (Image Signal Processor) and an ASC depth computing chip. The Aurora930 Pro depth camera uses structured-light 3D imaging technology. The Aurora930 Pro integrates an ISP (Image Signal Processor) and an ASC depth computing chip. Product Views ArmPi Ultra has a camera module mounted above the gripper. ArmPi Ultra uses a circular bearing assembly at the arm base. Black module with a perforated top grille secured to the robot's rear base platform ArmPi Ultra has a two-finger gripper. A camera module is mounted above the gripper on the arm. Specifications and Performance ArmPi Ultra uses a Hiwonder LX-225 bus servo at the highlighted lower arm joint. The highlighted LX-225 bus servo is labeled with a torque of 25 KG·CM; no voltage or test condition is specified. The robot's recognition software is described as using YOLOv8 with TensorRT optimization for real-time inference. The recognition software supports custom training, model-quantization deployment and cross-scenario transfer learning. The robot's recognition software is described as using YOLOv8 with TensorRT optimization for real-time inference. The recognition software supports custom training, model-quantization deployment and cross-scenario transfer learning. WonderEcho Pro measures 72 × 40 × 23 mm. WonderEcho Pro weighs 37 g. ArmPi Ultra uses six intelligent bus servos with real-time feedback and series-connected operation for coordinated multi-joint motion. ArmPi Ultra is stated to lift over 500 g. Application Examples ArmPi Ultra supports spoken interaction with verbal acknowledgment. A voice-control example requests a 3 cm upward movement and receives an acknowledgment. The Advanced Kit supports natural-language color selection using the instruction to keep objects matching the color of the sky and remove the rest. The Advanced Kit's illustrated color-sorting task identifies a blue block for transfer to a tray. ArmPi Ultra supports voice-directed sorting tasks, including keeping recyclable items and sorting other items into appropriate bins. The waste-sorting demonstration identifies objects labeled Ketchup, Rechargeable battery, and Bottle. ArmPi Ultra supports natural-language object selection and handover using shape and relative-size descriptions. The object-selection demonstration distinguishes a blue sphere, purple sphere, red cube, green cube, and blue cube. ArmPi Ultra supports spoken questions about its camera view and verbal descriptions of the scene. The scene-recognition demonstration identifies a Saiyan figure, a plant, and a white car model. ArmPi Ultra supports voice-directed object following using spatial descriptions, such as selecting the object above the milk. ArmPi Ultra supports voice-directed, color-based object placement, including placing a red object in a plate. ArmPi Ultra supports sequential object-placement instructions; the demonstrated sequence requests placement of a red object followed by a green object. ArmPi Ultra's 3D depth camera integrates RGB and depth data to capture color and depth point-cloud information. ArmPi Ultra combines RGB-D sensing with inverse kinematics for 3D spatial random grasping, sorting and transport. ArmPi Ultra is demonstrated in a conveyor-based waste-sorting setup, directing a block classified as plastic toward a blue bin. The robotic arm demonstrates grasping a beverage bottle with its gripper. ArmPi Ultra is demonstrated in a waste-sorting scenario with labeled blocks and colored bins. ArmPi Ultra is demonstrated gripping a screwdriver by its handle. ArmPi Ultra is presented for getting started with ROS; no ROS version is specified in this graphic. ArmPi Ultra is demonstrated grasping a red block with its gripper. ArmPi Ultra is presented in a shape-based object-moving demonstration with a response identifying a cuboid as the target. Additional Product Details Options and Model Comparison The 3D depth camera generates 3D images with depth information and retrieves RGB+D data. The 3D depth camera provides X, Y and Z coordinate data. The 3D depth camera generates 3D images with depth information and retrieves RGB+D data. The 3D depth camera provides X, Y and Z coordinate data. Interfaces and Connections The STM32 robotic arm controller accepts 7.4–8.4 V power input. The STM32 robotic arm controller has two LEDs.