Brain over Brawn: Using a Stereo Camera to Detect, Track, and Intercept a Faster UAV by Reconstructing the Intruder’s Trajectory

Abstract

This paper presents our approach to intercepting a faster intruder UAV, inspired by the MBZIRC 2020 Challenge 1. By utilizing a priori knowledge of the shape of the intruder’s trajectory, we can calculate an interception point. Target tracking is based on image processing by a YOLOv3 Tiny convolutional neural network, combined with depth calculation using a gimbal-mounted ZED Mini stereo camera. We use RGB and depth data from the camera, devising a noise-reducing histogram-filter to extract the target’s 3D position. Obtained 3D measurements of target’s position are used to calculate the position, orientation, and size of a figure-eight shaped trajectory, which we approximate using a Bernoulli lemniscate. Once the approximation is deemed sufficiently precise, as measured by the distance between observations and estimate, we calculate an interception point to position the interceptor UAV directly on the intruder’s path. Our method, which we have significantly improved based on the experience gathered during the MBZIRC competition, has been validated in simulation and through field experiments. Our results confirm that we have developed an efficient, visual-perception module that can extract information describing the intruder UAV’s motion with precision sufficient to support interception planning. In a majority of our simulated encounters, we can track and intercept a target that moves 30% faster than the interceptor. Corresponding tests in an unstructured environment yielded 9 out of 12 successful results.

Interception of an aerial manoeuvring target using monocular vision

“…As the problem statement in the paper is motivated by Challenge 1 of MBZIRC-2020, many other groups also worked in parallel to solve a similar problem [30,31,32]. In these algorithms, target depth information is used.…”

“…As R > 0, from ( 28), ( 29), (30), it is evident that the value of Vr , Vθ , and Vφ go to zero if the value of V θ and V φ go to zero. Therefore, any future value of V θ , V φ , and V r will remain the same if the value of V θ and V φ goes to zero.…”

“…In refs. [30,31], velocity command is generated proportional to tracking error between the current position of the interceptor and desired position, where the desired position is obtained after estimation of the target's 3D position. In ref.…”

“…The challenge involves grabbing a ball of diameter 100 mm hanging from a manoeuvring drone moving in a variable figure-of-eight trajectory in an unknown environment. Algorithms based on target position estimation are reported for interception in this scenario [30,31,32]. In refs.…”

“…Therefore, standard guidance algorithms cannot be applied directly using visual information. Existing approaches to vision-based interception are based on target pose estimation or error in the desired target pixel [11,26,27,28,30,31,32]; therefore, their efficiency will be limited in an unstructured outdoor environment. Also, interception strategies based on stereo vision will require higher computational requirements and payload budget.…”