The Control Method of Autonomous Flight Avoidance Barriers of UAVs in Confined Environments

Tiantian Dong, Yonghong Zhang, Qianyu Xiao, Yi ZhangSensors 2023 · DOI

TL;DR

This paper presents an enhanced 3D-Vector Field Histogram (3D-VFH) algorithm for autonomous flight and obstacle avoidance in confined environments for UAVs. By optimizing target point representation and employing a sliding window approach, the method achieves improved real-time performance and robust obstacle avoidance, even for dynamic obstacles.

What Problem Does This Paper Address?

Unmanned Aerial Vehicles (UAVs) are increasingly used for inspections and operations in dangerous or inaccessible areas. While outdoor UAV applications are maturing, operating in complex, confined environments presents significant challenges. These environments often lack GPS signals and adequate lighting, making traditional navigation methods unreliable. Furthermore, accurately perceiving and avoiding obstacles, especially dynamic ones, in such constrained spaces is crucial for safe and effective autonomous flight. Existing methods, while showing promise, often struggle with real-time performance and precise obstacle avoidance in these challenging scenarios.

Methods

The core of this work is an improved 3D-VFH algorithm. The authors introduce two key enhancements:

  • Polar Coordinate Target Point Transformation: Instead of using raw point cloud data, the algorithm converts target point coordinates into a polar coordinate system. This focuses on relevant local obstacle information, discarding long-range data that is less critical for immediate avoidance, thereby improving real-time performance.
  • Sliding Window for Optimal Path Selection: A sliding window algorithm is employed to estimate the optimal flight path. This method helps in selecting feasible flight directions and maintaining attitude stability during obstacle avoidance maneuvers. The paper specifically mentions a 4x4 sliding window, which samples more data points than a 3x3 window, potentially improving detection speed.

The paper also details the preprocessing of LiDAR point cloud data, including filtering and conversion to an OctoMap format for efficient processing. Coordinate transformations are performed to align the target point with the UAV's body coordinate system. The calculation of obstacle avoidance weights involves considering the distance to nodes and their occupancy.

Datasets

Not discussed in the paper. The experiments were conducted using a custom-built flight test platform and real-world scenarios.

Key Insights and Ideas

  • Focusing on Local Information is Key: The insight that long-range point cloud data is often irrelevant for immediate local obstacle avoidance in confined spaces is crucial. By transforming target point coordinates into a polar system, the algorithm effectively filters out noise and prioritizes relevant information, leading to better real-time performance.
  • Sliding Window for Robustness: The application of a sliding window for path selection is a clever way to ensure continuity and feasibility of flight directions, especially when dealing with dynamic obstacles. This approach helps in maintaining stable flight during avoidance maneuvers.
  • Integrated Approach for Confined Spaces: The paper demonstrates that combining advanced perception (LiDAR, visual cameras) with refined navigation algorithms like the improved 3D-VFH can enable robust autonomous flight in GPS-denied and low-light environments.

Key Results

The experimental results demonstrate the effectiveness of the proposed method:

  • Attitude Stability: The UAV exhibited good attitude stability during obstacle avoidance, with maximum pitch and roll angle errors within 2-3 degrees, meeting the requirements for confined environments.
  • Trajectory Following: The UAV was able to autonomously follow the expected trajectory and reach target points accurately.
  • Dynamic Obstacle Avoidance: The system successfully avoided a cylindrical obstacle placed in its path, proving its capability to handle dynamic obstacles not present in the global path planning.
  • Real-time Performance: The algorithm's real-time performance was improved due to the optimized target point representation. Obstacle detection was reported to occur within 0.5 seconds.
  • Altitude Stability: During fixed-point hovering, the actual visual height closely matched the expected height, with a maximum error of about 0.3 meters.

Limitations and Future Work

  • Movable Obstacle Testing: The authors acknowledge that the obstacle avoidance effect on movable obstacles has not been tested. They suggest adding random obstacles and modifying the algorithm based on their avoidance effect.
  • Target Point Overlap with Dynamic Obstacles: The paper notes that the current approach does not fully address situations where the target point itself might be a dynamic obstacle. This could lead to the UAV hovering near the obstacle instead of reaching the target.
  • Further Algorithm Modification: The authors suggest that the algorithm should be modified based on the obstacle avoidance effect, implying a need for continuous refinement.

Summary updated

Abstract

This paper proposes an improved 3D-Vector Field Histogram (3D-VFH) algorithm for autonomous flight and local obstacle avoidance of multi-rotor unmanned aerial vehicles (UAVs) in a confined environment. Firstly, the method employs a target point coordinate system based on polar coordinates to convert the point cloud data, considering that long-range point cloud information has no effect on local obstacle avoidance by UAVs. This enables UAVs to effectively utilize obstacle information for obstacle avoidance and improves the real-time performance of the algorithm. Secondly, a sliding window algorithm is used to estimate the optimal flight path of the UAV and implement obstacle avoidance control, thereby maintaining the attitude stability of the UAV during obstacle avoidance flight. Finally, experimental analysis is conducted, and the results show that the UAV has good attitude stability during obstacle avoidance flight, can autonomously follow the expected trajectory, and can avoid dynamic obstacles, achieving precise obstacle avoidance.
BibTeX
@inproceedings{dong2023control,
  author = {Tiantian Dong and Yonghong Zhang and Qianyu Xiao and Yi Zhang},
  title = {{The Control Method of Autonomous Flight Avoidance Barriers of UAVs in Confined Environments}},
  booktitle = {Sensors},
  year = {2023},
  doi = {10.3390/s23135896},
}
Figure