Challenges in Indoor Drone Inspection Applications
Indoor drone inspection has always faced key challenges, particularly in terms of positioning accuracy, flight stability, and spatial adaptability. These factors limit the effective application of conventional drones in complex indoor environments. The Amovlab SU17 drone breaks through these technical barriers by incorporating quad-camera VIO positioning, LiDAR SLAM, and LiDAR-based obstacle avoidance technologies. It effectively addresses issues such as mapping, positioning, path planning, and obstacle avoidance during indoor inspection flights, ensuring efficient operation in confined spaces.
Compared to other industrial drones, the Amovlab SU17 excels not only in stability but also in openness. As a highly integrated quadcopter designed for developers and industry users, the SU17 supports flexible secondary development interfaces, allowing for customized development based on industry needs. Its openness in both hardware and software endows it with strong scalability, making it adaptable to various complex inspection scenarios and capable of providing tailored solutions that significantly improve the efficiency and stability of indoor inspections.
Key Challenges to Address in Indoor Inspection
Indoor Positioning
Unlike outdoor environments, indoor inspections face more severe challenges as they cannot rely on GPS for positioning. Indoor environments require the integration of multiple sensors, such as IMUs, cameras, and LiDAR, for positioning.
Complex Obstacle Avoidance
Indoor environments typically contain numerous obstacles, requiring drones to have robust obstacle avoidance capabilities. Drones must be able to perceive their surroundings in real-time and dynamically adjust flight paths to ensure safe operation.
Indoor Map Interaction
In outdoor environments, platforms like Baidu Maps or Amap can be used to interactively plan waypoints. However, indoor environments generally lack maps. A key challenge is determining how to plan task waypoints for the drone to reach designated locations.
Physical Collision Protection
Indoor spaces are often confined, and the high-speed rotors of drones are prone to hitting walls or other obstacles. Lightweight protective structures are needed to ensure minor collisions do not cause flight anomalies.
Automated Takeoff and Landing
The core requirement of indoor inspection is eliminating the need for manual intervention. The drone must autonomously take off, land, recharge, and complete inspection tasks on a predefined schedule.
Amovlab SU17 Drone: Solving Indoor Inspection Challenges
The Amov SU17 drone provides comprehensive technical support for indoor inspections, particularly in positioning, obstacle avoidance, and mission execution. With its efficient hardware and intelligent algorithms, it demonstrates outstanding performance in complex indoor environments.
1. Solving Indoor Positioning Challenges
In this project, the SU17 drone relies on 3D LiDAR technology for precise positioning and mapping. Using the LiDAR SLAM (Simultaneous Localization and Mapping) algorithm, the SU17 can scan the surrounding environment in real-time and generate accurate grid maps. Even in GPS-denied indoor environments, the 3D LiDAR ensures stable flight and accurate positioning.
The LiDAR system scans the 3D structure of the indoor environment to perceive obstacles in real-time, providing a solid foundation for path planning and obstacle avoidance. This technology performs exceptionally well in complex environments, particularly in confined spaces with numerous obstacles.
The 3D LiDAR SLAM employs the FAST-LIO algorithm, which excels in 3D LiDAR SLAM scenarios. However, it may face challenges, such as divergence, in environments with glass or unclear geometric features. Future development focuses on the fusion of vision, LiDAR, IMU, and GPS to enhance adaptability. Amov Lab has introduced the BSA-SLAM algorithm library and plans to launch a fusion system of LiDAR, vision, and GPS by mid-2025 to further improve system robustness.
2. Real-Time Obstacle Avoidance and Dynamic Path Planning
During flight, the SU17’s 3D LiDAR continuously perceives the surrounding environment, automatically identifying and avoiding obstacles. Using the LiDAR SLAM algorithm, the drone dynamically adjusts its flight path based on environmental changes, ensuring smooth task execution while avoiding potential collisions. Additionally, a depth map generated by the quad-camera vision system can be used for mapping, obstacle avoidance, and path planning. The obstacle avoidance algorithm, ego-swarm, offers excellent dynamic obstacle avoidance performance, effectively avoiding moving obstacles encountered along flight routes. This algorithm is integrated into the open-source Prometheus motion planning project, which users can simulate.
Project URL:
https://gitee.com/amovlab/Prometheus
3. Digital Twin Mapping and Waypoint Planning
To optimize flight tasks for indoor inspections, STL-format digital twin maps are imported into the ground control station. Users can use these maps to plan waypoints, set flight paths, and adjust tasks automatically. Before actual flights, users can visualize the entire environment’s 3D model through the ground station and plan task routes accordingly. Digital twin maps can be constructed in various ways; the ideal method involves constructing maps during drone flights. Alternatively, data can be collected in advance—combining LiDAR and RGB data for post-processing and fusion. In the demonstration video, an STL-format 3D map constructed in advance is shown, without using onboard LiDAR or camera data for mapping.
4. Task Execution
After flight path planning, the SU17 drone executes a series of pre-defined tasks, such as capturing photos, recording videos, or monitoring the environment. The dwell time at each task point can be flexibly adjusted to ensure high-quality data collection. With precise path planning and efficient task execution, the SU17 ensures that every inspection mission is accurately completed. The ASDK interface on the SU17 hardware platform allows the retrieval of any onboard sensor data, such as camera gimbal video streams and tilt angle control.
5. Visual Recognition and Vision-Guided Landing
Upon completing inspection tasks, the SU17 drone automatically returns to its starting point and uses vision-based precision-guided landing algorithms to land at designated locations. When equipped with a wireless charging module, the SU17 can also recharge automatically, further improving endurance and preparing for the next inspection task. Vision guidance capabilities are supported by our open-source SpireCV algorithm library.
Project URL:
https://gitee.com/amovlab/SpireCV
With the aforementioned software framework and the SU17 hardware’s expandable and developer-friendly features, a complete indoor inspection solution can be built in a short time. The development process is illustrated below:
Openness and Developer-Friendliness
One of the key differentiating features of the Amov SU17 drone is its strong openness compared to most indoor inspection robots on the market. Rather than being a closed system, the SU17 is a highly integrated and open drone development platform, designed specifically for drone developers and industry users to enable deep secondary development.
Hardware and Software Platform Openness:
The SU17 integrates state-of-the-art hardware, including quad-camera SLAM, RTK, 3D LiDAR, and optical flow altitude sensors, all of which can be customized and expanded based on needs. Additionally, the SU17 is equipped with self-developed flight control systems and a high-performance onboard computer, meeting diverse development requirements for various application scenarios.
Support for Secondary Development:
Amov provides users with a complete secondary development toolkit. Through ASDK-G (ground station application development interface) and ASDK-D (drone body data retrieval interface), users can develop custom functions, such as gimbal control, video streaming, IMU data collection, and flight route planning, fully meeting the specialized needs of scientific research and industry.
Professional Ground Control Station:
The SU17 is equipped with a professional ground control station system developed on a cross-platform QT framework. Users can obtain secondary development authorization to create a customized ground station system tailored to their industry and functional requirements. This open design makes the SU17 adaptable to various application scenarios, offering solutions ranging from research to industrial applications.
Future Vision: Fusion of Visual SLAM and LiDAR SLAM
Although the current project relies on 3D LiDAR for positioning, Amov is actively developing a visual SLAM and LiDAR SLAM fusion positioning system. This innovative technology will further enhance the SU17’s performance in indoor inspections. By integrating visual and LiDAR data, the drone will achieve more precise positioning and obstacle avoidance.
The combination of visual SLAM and LiDAR SLAM technologies will enable the SU17 to navigate and execute tasks more efficiently in dynamic and complex environments, opening up broader prospects for indoor inspection applications.
Conclusion
Whether in factories, warehouses, or other complex indoor environments, the Amov SU17 can reliably and efficiently complete tasks, providing robust technical support for indoor inspection. Its open hardware platform and secondary development interfaces make the SU17 not just a drone system but a customizable and expandable development platform.
We look forward to collaborating with you to explore more intelligent inspection applications and make your work more efficient and smarter.