Revisiting Stereo Triangulation in UAV Distance Estimation

TL;DR

This work builds and presents a UAVDE dataset for UAV distance estimation, in which distance between two UAVs is obtained by UWB sensors, and proposes a novel position correction module that can directly predict the offset between the observed positions and the actual ones and then perform compensation in stereo triangulation calculation.

Abstract

Distance estimation plays an important role for path planning and collision avoidance of swarm UAVs. However, the lack of annotated data seriously hinders the related studies. In this work, we build and present a UAVDE dataset for UAV distance estimation, in which distance between two UAVs is obtained by UWB sensors. During experiments, we surprisingly observe that the stereo triangulation cannot stand for UAV scenes. The core reason is the position deviation issue due to long shooting distance and camera vibration, which is common in UAV scenes. To tackle this issue, we propose a novel position correction module, which can directly predict the offset between the observed positions and the actual ones and then perform compensation in stereo triangulation calculation. Besides, to further boost performance on hard samples, we propose a dynamic iterative correction mechanism, which is composed of multiple stacked PCMs and a gating mechanism to adaptively determine whether further correction is required according to the difficulty of data samples. We conduct extensive experiments on UAVDE, and our method can achieve a significant performance improvement over a strong baseline (by reducing the relative difference from 49.4% to 9.8%), which demonstrates its effectiveness and superiority. The code and dataset are available at https://github.com/duanyuan13/PCM.

Figures (10)

• Figure 1: Samples from our presented UAVDE dataset. We collect thousands of UAV stereo images and annotate them with UAV bounding boxes and distances to the UAV center. Notably, only distance to the UAV center is collected via UWB sensors instead of collecting pixel-wise annotations on LiDAR, which is more economical and efficient for practical UAV applications. Best viewed in color.
• Figure 2: Stereo triangulation in UAV scenes. The center of UAVs in stereo images can be obtained by UAV detection, based on which stereo triangulation can be computed to estimate the UAV distance. Here, $B$ and $f$ are the baseline and focal length of the stereo camera, respectively.
• Figure 3: Upper-bound analysis of stereo triangulation in UAV scenes. The prediction is relatively accurate when the distance is small (e.g., less than 8 m), but as the distance increases, the prediction gradually deviates from the ground truth.
• Figure 4: The illustration of the recording UAV. The UAV is equipped with a stereo camera for image data collection and a UWB sensor for distance annotation collection.
• Figure 5: The illustration of typical UAV scenarios. We collect data in typical UAV scenarios, e.g., playground, forest, building and basketball court.