Vishal_Singh's open-source robot integrates SLAM, vision, and voice on Raspberry Pi

Vishal_Singh has introduced an open-source autonomous navigation and interactive robot project that aims to combine mobility, perception, and natural human-robot interaction on a single platform. The hardware is built around a Raspberry Pi, an RP LiDAR sensor for mapping and obstacle avoidance, a C1 camera for computer vision, and an Android tablet interface for user interaction. The robot also includes various robotics sensors and a motor control system. On the software side, the stack is heavily ROS2-based, integrating MediaPipe for face and gesture recognition, RTAB-Map for SLAM, FastAPI for cloud communication, Docker for containerization, and Whisper for speech-to-text. To simulate and test digital twins, the project leverages NVIDIA Isaac Sim.

A key aspect of this design is the hybrid AI architecture: computationally intensive perception and AI workloads—such as advanced object detection or natural language understanding—are offloaded to cloud-based NVIDIA H200 GPUs, while real-time control (e.g., motor commands, sensor fusion) remains on the Raspberry Pi for low latency. Planned features include full autonomous navigation with LiDAR-based SLAM, human detection and tracking, gesture recognition, real-time telemetry, and a voice assistant capable of natural language interaction. Currently, the development is focused on ROS2 integration, the navigation stack, perception pipelines, and human-robot interaction (HRI). The project is hosted publicly on GitHub for community contributions and feedback on architecture and ROS2 best practices.