UAV Navigation in Tunnels with 2D tilted LiDARs

Danilo Tardioli,Lorenzo Cano,Alejandro R. Mosteo

2024-04-15

Abstract:Navigation of UAVs in challenging environments like tunnels or mines, where it is not possible to use GNSS methods to self-localize, illumination may be uneven or nonexistent, and wall features are likely to be scarce, is a complex task, especially if the navigation has to be done at high speed. In this paper we propose a novel proof-of-concept navigation technique for UAVs based on the use of LiDAR information through the joint use of geometric and machine-learning algorithms. The perceived information is processed by a deep neural network to establish the yaw of the UAV with respect to the tunnel's longitudinal axis, in order to adjust the direction of navigation. Additionally, a geometric method is used to compute the safest location inside the tunnel (i.e. the one that maximizes the distance to the closest obstacle). This information proves to be sufficient for simple yet effective navigation in straight and curved tunnels.

Robotics

What problem does this paper attempt to address?

The paper attempts to address the problem of how Unmanned Aerial Vehicles (UAVs) can achieve high-speed autonomous navigation in complex environments such as tunnels or mines, where GNSS signals are unavailable, lighting is insufficient or non-existent, and wall features are sparse. Specifically, the authors propose a novel navigation technique based on 2D inclined LiDAR information, using a combination of geometric and machine learning algorithms to solve this problem.

### Main Contributions of the Paper:
1. **No Dependence on Precise Localization**: Unlike traditional navigation methods that require precise absolute localization, this method does not require the UAV to perform precise self-localization. Instead, it achieves navigation through relative position and direction adjustments.
2. **Use of 2D Inclined LiDAR**: By installing two 2D inclined LiDARs (one inclined upwards and one inclined downwards), point cloud data of the tunnel walls is obtained to estimate the yaw angle of the UAV relative to the longitudinal axis of the tunnel.
3. **Deep Neural Networks**: Deep neural networks are used to process the LiDAR data to estimate the UAV's yaw angle, thereby adjusting its direction to align with the longitudinal axis of the tunnel.
4. **Geometric Methods**: Geometric methods are used to calculate the safest position within the tunnel, i.e., the position farthest from the nearest obstacle, ensuring that the UAV does not collide with the walls during navigation.
5. **Applicable to Both Straight and Curved Tunnels**: This method is applicable not only to straight tunnels but also to curved tunnels, maintaining the safety of the UAV in simple and effective navigation.

### Specific Problems Addressed:
- **Navigation in GNSS-Denied Environments**: In underground environments, GNSS signals are unavailable, rendering traditional navigation methods ineffective.
- **Insufficient or Non-Existent Lighting**: Poor lighting conditions in underground environments make it difficult for visual sensors to function.
- **Sparse Wall Features**: The lack of prominent feature points inside tunnels makes feature-based navigation methods difficult to apply.
- **High-Speed Navigation**: Achieving high-speed autonomous navigation within tunnels while ensuring safety.

Through these innovations, the paper provides an efficient and safe navigation solution for complex underground environments.