Monocular Vision-based Sense and Avoid of UAV Using Nonlinear Model Predictive Control

Abstract

SummaryThe potential use of onboard vision sensors (e.g., cameras) has long been recognized for the Sense and Avoid (SAA) of unmanned aerial vehicles (UAVs), especially for micro UAVs with limited payload capacity. However, vision-based SAA for UAVs is extremely challenging because vision sensors usually have limitations on accurate distance information measuring. In this paper, we propose a monocular vision-based UAV SAA approach. Within the approach, the host UAV can accurately and efficiently avoid a noncooperative intruder only through angle measurements and perform maneuvers for optimal tradeoff among target motion estimation, intruder avoidance, and trajectory tracking. We realize this feature by explicitly integrating a target tracking filter into a nonlinear model predictive controller. The effectiveness of the proposed approach is verified through extensive simulations.

Vision-Based Mid-Air Object Detection and Avoidance Approach for Small Unmanned Aerial Vehicles with Deep Learning and Risk Assessment

“…It controls the host aircraft based on the current state and solves the collision through local motion control. One planning avoidance method utilizes nonlinear model predictive control (NMPC) to avoid an intruder with a single vision sensor [16]. Another study [17] utilized a doppler radar to implement collision avoidance.…”

Sensor-Model-Based Trajectory Optimization for UAVs to Enhance Detection Performance: An Optimal Control Approach and Experimental Results

“…In this work, we reuse the two optimization methods nonlinear model predictive control (NMPC) and dynamic programming & optimal control (DP&OC), which were proposed in our previous works [ 13 , 14 ]. NMPC is a well-established method for UAV trajectory optimization and is used for example by [ 15 , 16 , 17 , 18 ]. For the application of DP&OC for path and trajectory planning, we refer to [ 19 , 20 , 21 , 22 ].…”

Risk continuity analysis of transmission line channel based on historical hidden danger data

“…Because the visual inspection of transmission line channels is widely used, the automatic identification of visual information has been realized and the alarm objects appear in the image, such as machinery, fireworks, foreign objects, etc. In addition to the basic statistical analysis report, can be based on the alarm data for data mining, such as continuous alarm of real-time alarm data judgment, alarm level based on the results of intelligent annotation, image recognition model suspected false alarm and omission sample identification, but the application of the above scenarios need continuous alarm identification technical support, fireworks, foreign body alarm for low frequency, seasonal periodic, regular single comparison and mechanical have good applicable dimension analysis can not effectively identify the fireworks, foreign body alarm [1] .…”