Hardware & physical systems
Real machines, real sensors, real feedback. Cogninoid Labs builds hardware that can be touched, measured, and improved — not just simulated.
Adeept Modular Arm v1
In ProgressA 6-axis robotic arm based on the Adeept ADA031 platform, modified for AI-driven control experiments. Firmware in C++/Arduino; Python control API in development. Integrated into the Cogninoid device ecosystem.
✦ First hardware platform integrated into the MEIDLab AI agent system.
FDM Robotic Gripper Mk1
In ProgressA 3D-printed adaptive gripper designed for the Adeept arm. Uses a rack-and-pinion mechanism driven by a servo, with a force-sensitive resistor for grasp detection. Designed parametrically in Fusion 360.
✦ Fully parametric design — any jaw width or finger length can be regenerated in minutes.
Sensor Fusion Node
PlannedA lightweight embedded module combining IMU, force/torque, and proximity sensors into a unified ROS node. Designed to be strapped to any robot link for real-time state monitoring.
✦ Modular hardware design — swappable sensor payloads without firmware changes.
3D Print Farm Controller
In ProgressA Python-based orchestration system for managing multiple FDM 3D printers in a small lab setting — queue management, failure detection, and print job logging.
✦ Reduces manual monitoring overhead for long print jobs by 80%.
High-Speed Vision Module
PlannedA compact vision module using a global-shutter camera for real-time object detection and pose estimation in robotic manipulation tasks. Designed to mount to any standard robot wrist.
✦ Global shutter eliminates motion blur artifacts during fast arm movements.
Autonomous Lab Bench
PlannedA self-contained tabletop platform where an AI agent can autonomously run small-scale physical experiments — moving samples, actuating valves, measuring properties, and logging results.
✦ Designed to run overnight experiments without human supervision.
Rapid Prototyping Jig Kit
In ProgressA library of parametric jig and fixture designs for common lab tasks: PCB holders, sensor mounts, cable guides, and alignment tools. Generated via OpenSCAD with configurable parameters.
✦ Reduces custom fixture design time from hours to minutes.
Embedded AI Inference Node
PlannedA compact compute module based on Raspberry Pi + Hailo-8 accelerator, designed to run neural network inference at the robot edge — decision latency under 20ms.
✦ Edge inference eliminates cloud round-trip latency for real-time control tasks.