LSR550 2WD Differential-Drive ROS Research Robot Platform

AI, Perception & ROS Development Features

The LSR550 2WD development library combines ROS navigation, perception, AI integration, interaction, and optional outdoor-autonomy workflows. Voice-dependent functions require the corresponding voice hardware, while local AI-model workloads depend on the selected onboard computer and model requirements.

01 AI Large-Model Integration

Large Language Model Integration

Supports online LLM integration workflows for natural-language command interpretation, dialogue, and higher-level task interaction.

Voice-Model Interaction

With the voice module and speaker, speech-to-text and text-to-speech workflows can be combined with AI-model interaction.

Vision-Language Interaction

Camera-based workflows can analyze visual content, recognize object categories, and return text or voice responses through compatible multimodal models.

Local Model Deployment

Compatible onboard-compute configurations support local AI-model deployment for lower-latency inference and reduced dependence on cloud services.

02 AI Multimodal Applications

AI Vision: Scene Understanding & Perception

Multimodal vision workflows can analyze images or video, identify scene elements, and generate semantic feedback.

AI Recognition: Object Tracking

Vision-model workflows support target identification and real-time tracking experiments.

AI Interaction: Color-Aware Line Following

Multimodal interaction can connect color-based commands with vision line-following behaviors.

Flexible Chassis Control

AI interaction workflows can be connected to chassis motion commands, with LiDAR-based emergency-stop behavior for mobile-robot experiments.

AI Interaction: LiDAR Tracking

Natural-language interaction can be linked to LiDAR-based tracking behaviors for target-following experiments.

RAG + Dual-Model Inference Workflow

RAG workflows retrieve knowledge-base context for task interpretation, while language and multimodal models can be combined for task planning and perception-driven actions.

Voice Wake-Up & Interruption

With the voice module, new voice commands can interrupt an ongoing interaction workflow and trigger a new dialogue or task request.

03 Embodied AI SLAM Mapping & Semantic Navigation

Semantic Mapping

Multimodal interaction can be connected to mapping commands so users can initiate mapping and save maps through semantic task workflows.

Semantic Navigation

Semantic coordinate labels can be associated with map locations so navigation requests can be translated into target points.

Voice Semantic Labeling

With the voice module, semantic location labels can be added through dialogue during navigation workflows.

Combined Embodied-AI Workflow

Preconfigured interaction nodes can connect voice, perception, navigation, and mobile-robot functions for integrated embodied-AI demonstrations and secondary development.

04 High-Precision Motion Sensing

500-Line GMR Encoder Feedback

The drivetrain uses 500-line AB-phase GMR encoders for wheel-speed and odometry feedback, supporting low-speed navigation and precise maneuvering.

05 Outdoor Navigation & 3D LiDAR Perception (Optional)

Multi-Line LiDAR Integration

An optional 16-line mechanical LiDAR uses TOF ranging and provides 360° horizontal perception for outdoor point-cloud mapping and autonomous-driving research. The platform feature set specifies measurement precision within ±3 cm for this 16-line LiDAR workflow.

Autoware Outdoor Mapping & Navigation

Autoware-oriented workflows support HD-map-based path planning for low-speed autonomous-navigation research.

Point-Cloud Obstacle Avoidance

LiDAR point-cloud clustering can be used for obstacle perception and avoidance in outdoor navigation experiments.

The Ultra configuration includes multi-line LiDAR as standard. NVIDIA Orin controllers are suitable for higher-compute Autoware and 3D-perception workloads.

06 Autonomous Power Management (Optional)

Autonomous Docking & Recharging

Auto-recharging configurations support return-to-station and docking workflows using ROS mapping and navigation. Development resources include charging-station calibration, docking methods, and low-battery return logic.

Open-Source ROS Development Library

The following ROS, perception, navigation, AI, and control examples are included in the development library.

1. RTAB-Map 3D VSLAM & LiDAR Integration

Supports RTAB-Map vision mapping and LiDAR–vision fusion workflows for 3D environment reconstruction in ROS 1 and ROS 2.

2. Classic 2D LiDAR Mapping & Navigation

ROS 2 resources include Gmapping, Cartographer, and slam_toolbox. ROS 1 resources include Gmapping, Hector, Karto, and Cartographer, together with point-to-point, waypoint, and obstacle-avoidance workflows.

3. ORB-SLAM2 Visual Mapping

Uses ORB-SLAM2 for camera pose estimation and sparse 3D reconstruction, with metric-scale information available in RGB-D workflows.

4. ROS Qt Graphical User Interface

A Qt-based interface provides convenient ROS launch controls and real-time feedback such as robot velocity, battery status, and system information.

5. YOLO Object Detection

ROS 1 resources include YOLOv3 workflows, while ROS 2 resources include YOLOv8 and YOLO11 workflows for object detection and custom-model experiments.

6. LLM Deployment

Supports online LLM integration and local-model workflows on compatible onboard-compute configurations for reasoning, dialogue, and robot interaction.

7. Depth-Based Visual Following

Uses RGB-D depth information to estimate target distance and bearing for real-time visual-following experiments.

8. KCF Target Tracking

Includes Kernelized Correlation Filter tracking for identifying and following targets with stable visual features.

9. AR Tag Recognition & Following

Supports 6-DOF AR-tag pose detection for tag following and tag-based localization experiments.

10. RRT Autonomous Exploration

Includes an RRT-based autonomous exploration workflow for mapping, map saving, and return-to-start experiments.

11. Web-Based Camera Monitoring

Provides browser-based access to the robot camera stream for remote monitoring and development.

12. RGB Camera Line Following

Supports RGB-vision line following with LiDAR-assisted obstacle avoidance.

13. LiDAR-Based Following

Uses LiDAR scans for nearby-target selection and following experiments.

14. LiDAR Angle Masking

SDK-based LiDAR angle masking allows selected scan sectors to be filtered according to the sensor and application configuration.

15. Sound Source Localization

With a compatible microphone array, sound-direction estimation supports voice-interaction and navigation experiments with 1° directional precision.

16. Voice-Activated Summoning

With the voice module, microphone-array direction information and LiDAR ranging can be combined in voice-driven summoning experiments.

17. Voice-Controlled Navigation

Voice-command workflows can associate recognized commands with predefined navigation points on a saved map.

18. TTS Interaction

Text-to-speech workflows support robot voice interaction and can be extended with compatible online voice-dialogue services.

19. TEB & DWA Path Planning

Includes tutorials and Python-based learning resources for TEB and DWA navigation and local path-planning methods.

20. DWB / MPPI / RPP Controllers

Provides DWB, MPPI, and Regulated Pure Pursuit controller examples for different Nav2 scenarios and robot footprints.

21. Full-Coverage Path Planning

Supports generation of coverage paths for user-defined areas, useful for inspection and systematic-area traversal research.

22. Chassis Kinematics Analysis

Development resources include kinematics examples for Ackermann, differential, tracked, Mecanum, omnidirectional, and 4WD chassis types.

23. Path Recording & Playback

Manual trajectories can be recorded and replayed autonomously using Nav2-based workflows.

24. URDF Models

Includes URDF models for simulation, visualization, and robot-software development based on platform geometry.

25. ROS Mobile App for Navigation

A mobile application supports robot motion control, mapping, and navigation workflows.

26. Hardware Tuning & Parameter App

Android and iOS tools support parameter tuning, gravity-sensing control, and waveform visualization for supported hardware functions.

27. Wireless Code Flashing

Bluetooth tools support wireless code flashing and remote debugging for supported controller workflows.

28. Optional 4G/5G Remote Modules

Optional cellular communication modules support long-range video transmission and remote-control development over compatible mobile networks.

29. 3D Reconstruction for Autonomous-Driving Research

With optional multi-line LiDAR, the platform supports outdoor 3D mapping and environment-reconstruction workflows for autonomous-driving research.

30. Autoware.Universe Multi-Point Navigation

Autoware.Universe examples include sequential waypoint dispatch using Python scripts for continuous multi-point navigation experiments.

Frequently Asked Questions

What is the LSR550 2WD designed for?

The LSR550 2WD is designed for ROS research, SLAM, autonomous navigation, computer vision, embodied AI, path planning, mobile-robot control, and secondary development.

What are the main chassis specifications?

The 2WD differential-drive chassis has a 30 kg rated payload, 1.3 m/s maximum speed, 9.4 kg net weight, 408 × 380.7 × 274.6 mm dimensions, 35° maximum climbing incline, and approximately 9 hours of no-load endurance.

What motors and encoders are used?

The drivetrain uses MD36L 60 W brushed DC motors with a 1:27 reduction ratio and 500-line AB-phase GMR encoders for wheel-speed and odometry feedback.

Which onboard computers are available?

Controller options include Raspberry Pi 5 (8 GB), Jetson Orin Nano Super (8 GB), and Jetson Orin NX Super (16 GB). A Basic Package without the ROS computer, LiDAR, and RGB-D depth-camera bundle is also available on Standard configurations.

Which ROS versions are supported?

Raspberry Pi 5 supports ROS 1 Melodic and ROS 2 Humble. Jetson Orin Nano Super and Jetson Orin NX Super support ROS 1 Noetic and ROS 2 Humble.

What 2D LiDAR options are available?

N10P and M10P TOF LiDAR options are available. N10P provides a 25 m detection radius with adjustable 6–12 Hz scanning, while M10P provides a 30 m detection radius with 12 Hz scanning. Both provide 360° scanning.

Does the robot include an RGB-D depth camera?

ROS controller packages include an RGB-D depth camera with 0.6–8 m depth range, 640 × 480 depth output at 30 fps, and 1920 × 1080 RGB output at 30 fps. The camera is specified for indoor operation.

Does the LSR550 2WD support 3D LiDAR?

Yes. The Ultra configuration includes multi-line LiDAR as standard, supporting 360° 3D perception, point-cloud mapping, obstacle perception, and outdoor autonomous-navigation research.

Can I use Autoware with this platform?

Yes. Autoware-oriented workflows include HD-map-based path planning, point-cloud obstacle avoidance, 3D reconstruction, and multi-point navigation. NVIDIA Orin controllers and multi-line LiDAR are suitable for these workloads.

What SLAM and navigation resources are included?

Development resources include RTAB-Map, Gmapping, Cartographer, slam_toolbox, ORB-SLAM2, RRT autonomous exploration, Nav2, TEB, DWA, DWB, MPPI, RPP, coverage planning, path recording and playback, and waypoint navigation examples.

Does it support AI and large-model applications?

Yes. The development library includes LLM integration, multimodal vision workflows, RAG-oriented task interaction, semantic mapping and navigation, object detection, target tracking, and local-model workflows on compatible onboard-compute configurations.

Which functions require the voice module?

Voice-driven functions such as speech interaction, voice wake-up and interruption, sound-source localization, voice-controlled navigation, voice semantic labeling, and voice-driven summoning require the corresponding voice hardware.

Does the robot support autonomous charging?

Yes. Auto-recharging configurations support return-to-station and docking workflows using ROS mapping and navigation, including charging-station calibration and low-battery return logic.

Can I add 4G or 5G remote communication?

Yes. Optional 4G / 5G communication modules can be used for cellular video transmission and remote-control development over compatible mobile networks.

What development resources are included?

Development resources include ROS source code and system images, navigation and perception examples, URDF models, tutorials, mobile tools, controller examples, path-planning workflows, AI integration examples, and secondary-development resources.