rs_perception is the official LiDAR perception package by RoboSense Technology. Full pipeline: point cloud segmentation, 3D bounding box detection, classification (pedestrians, vehicles, cyclists/motorcycles, others) and multi-object tracking. Intended for AV and mobile robotics.
SDK uses rs_driver as data source. Pipeline: (1) preprocessing (filtering, voxelisation), (2) ground removal, (3) object cluster segmentation, (4) DNN classification (CPU + GPU/CUDA), (5) object tracking (Kalman + Hungarian). Results: object lists with ID, type, 3D bounding box, velocity, heading.
Official ROS 2 node publishing results as rsperception_msgs. Supports ROS 2 Humble and Iron (Ubuntu 22.04). Online and offline (.pcap / ROS 2 bags) modes. Compatible with Nav2, Autoware Universe.
CPU: x86_64/ARM64, 4 cores, 4 GB RAM. GPU: NVIDIA CUDA 11.x/12.x, 6 GB VRAM; TensorRT optional on Jetson Orin/AGX Xavier.
Dual-license: open-source Apache 2.0 (GitHub) + proprietary commercial version with advanced DL models. Build: CMake + colcon; binary DL models downloaded after registration on the RoboSense portal.
An SDK (Software Development Kit) is a curated set of libraries, interfaces, tools, sample code, and documentation intended for building applications and integrating with a specific hardware device, platform, or service. In robotics, an SDK typically exposes device control, telemetry, sensor access, configuration, and execution functions, significantly reducing the time-to-first-integration for developers targeting a specific robot or platform.
Used in AV vehicles with RoboSense LiDARs (BYD, XPeng, GAC) and in AGV/AMR integrators. Compatible with Autoware Universe.
GitHub: ~200+ stars (RoboSense-LiDAR/rs_perception)
Additional 2–5 GB for binary DNN models (downloaded after registration on the RoboSense portal).
License family: Permissive
License family: Proprietary – Commercial
