Energy-Aware Design of Vision-Based Autonomous Tracking and Landing of a UAV

Abstract

In this paper, we present the design and evaluation of a vision-based algorithm for autonomous tracking and landing on a moving platform in varying environmental conditions. We use an energy-aware approach, where the design of the algorithm is based on an evaluation of the energy consumption and Quality of Service (QoS) of each computational component. We evaluate our approach with an agricultural use case where a moving platform is tracked using a landing marker and the YOLOv3tiny CNN is used to detect ground-based hazards. We perform all computations onboard using an NVIDIA Jetson Nano and analyse the impact on the flight time by profiling the energy consumption of the marker detection and the CNN. Experiments are conducted in Gazebo simulation using an energy modeling tool to measure the computational energy cost as a function of QoS. We test the energy efficiency and robustness of our system in various dynamic wind disturbances. We show that the marker detection algorithm can be run at the highest QoS with only a marginal energy overhead whereas adapting the QoS level of CNN results in a considerable power saving. The power saving is significant for a system executing on a fixed-wing UAV.

2023 Design, Automation & Test in Europe Conference & Exhibition (DATE)

“…The proper functioning of the use cases depend on correct soft-timely execution and low energy consumption. Due to overlapping frames, for instance, missing a few deadlines is possible without missing objects, whereas energy constraints are tighter, directly impacting the flight time and thus coverage area [33].…”

“…Three different hardware platforms have been tested: 1) an Apalis TK1; 2) Nvidia TX2; and, 3) Nvidia Nano. In all cases, Linux OS runs use case-specific software, e.g., C++/CUDA computer vision algorithms for detections on the GPU, and is eventually complemented with Robot Operating System (ROS) middleware [31], [33]. We applied the TeamPlay toolchain for unpredictable architectures but, as a result of platform complexity, omitted fact checker and SecurityAnalyser, which both require further investigation.…”

“…Using our TeamPlay methodology, we observe an energy improvement of 18%, resulting in the flight time being increased by approximately 4 minutes. Conversely to the SAR use case, the PA use case [31]- [33] utilized merely energy analysis, yet enabled flight time optimisations. When cruising, the mechanical components of the UAV consumed 28 Watts on average, whereas software components consumed between 2 and 11 Watts, with the toolchain enabling in-flight battery-aware schedulability [31].…”

Accurate Calibration of Power Measurements from Internal Power Sensors on NVIDIA Jetson Devices

“…Energy consumption and efficiency have always been crucial aspects for energy-constrained and battery-powered devices in network edge scenarios [3]. The authors of [4] proposed an energy-aware autonomous tracking and landing system for a Unmanned Aerial Vehicle (UAV) equipped with an NVIDIA Jetson nano, exploiting the Jetson onboard power sensors to improve their solution. However, there have been debates [5] over the measurement methods and the precision of NVIDIA Jetson built-in sensors.…”

Energy-Aware Planning-Scheduling for Autonomous Aerial Robots

“…Our focus is on fixed wings, i.e., airborne robots where wings provide lift, propellers provide forward thrust, and control surfaces perform maneuvering. Here, motion and computations energies are within an order of magnitude from each other [25], [26]. There are other classes where planningscheduling energy awareness leads to irrelevant savings, i.e., when the motion energy contribution far outreaches the computations or vice-versa.…”