Showing 1–6 of 6 results for author: Kyas, M

Lowering Barriers to Entry for Fully-Integrated Custom Payloads on a DJI Matrice

Authors: Joshua Springer, Gylfi Þór Guðmundsson, Marcel Kyas

Abstract: Consumer-grade drones have become effective multimedia collection tools, spring-boarded by rapid development in embedded CPUs, GPUs, and cameras. They are best known for their ability to cheaply collect high-quality aerial video, 3D terrain scans, infrared imagery, etc., with respect to manned aircraft. However, users can also create and attach custom sensors, actuators, or computers, so the drone… ▽ More Consumer-grade drones have become effective multimedia collection tools, spring-boarded by rapid development in embedded CPUs, GPUs, and cameras. They are best known for their ability to cheaply collect high-quality aerial video, 3D terrain scans, infrared imagery, etc., with respect to manned aircraft. However, users can also create and attach custom sensors, actuators, or computers, so the drone can collect different data, generate composite data, or interact intelligently with its environment, e.g., autonomously changing behavior to land in a safe way, or choosing further data collection sites. Unfortunately, developing custom payloads is prohibitively difficult for many researchers outside of engineering. We provide guidelines for how to create a sophisticated computational payload that integrates a Raspberry Pi 5 into a DJI Matrice 350. The payload fits into the Matrice's case like a typical DJI payload (but is much cheaper), is easy to build and expand (3D-printed), uses the drone's power and telemetry, can control the drone and its other payloads, can access the drone's sensors and camera feeds, and can process video and stream it to the operator via the controller in real time. We describe the difficulties and proprietary quirks we encountered, how we worked through them, and provide setup scripts and a known-working configuration for others to use. △ Less

Submitted 9 May, 2024; originally announced May 2024.

Comments: 5 pages, 3 figures, 1 table, workshop paper, version 1 (preprint)

A Precision Drone Landing System using Visual and IR Fiducial Markers and a Multi-Payload Camera

Authors: Joshua Springer, Gylfi Þór Guðmundsson, Marcel Kyas

Abstract: We propose a method for autonomous precision drone landing with fiducial markers and a gimbal-mounted, multi-payload camera with wide-angle, zoom, and IR sensors. The method has minimal data requirements; it depends primarily on the direction from the drone to the landing pad, enabling it to switch dynamically between the camera's different sensors and zoom factors, and minimizing auxiliary sensor… ▽ More We propose a method for autonomous precision drone landing with fiducial markers and a gimbal-mounted, multi-payload camera with wide-angle, zoom, and IR sensors. The method has minimal data requirements; it depends primarily on the direction from the drone to the landing pad, enabling it to switch dynamically between the camera's different sensors and zoom factors, and minimizing auxiliary sensor requirements. It eliminates the need for data such as altitude above ground level, straight-line distance to the landing pad, fiducial marker size, and 6 DoF marker pose (of which the orientation is problematic). We leverage the zoom and wide-angle cameras, as well as visual April Tag fiducial markers to conduct successful precision landings from much longer distances than in previous work (168m horizontal distance, 102m altitude). We use two types of April Tags in the IR spectrum - active and passive - for precision landing both at daytime and nighttime, instead of simple IR beacons used in most previous work. The active IR landing pad is heated; the novel, passive one is unpowered, at ambient temperature, and depends on its high reflectivity and an IR differential between the ground and the sky. Finally, we propose a high-level control policy to manage initial search for the landing pad and subsequent searches if it is lost - not addressed in previous work. The method demonstrates successful landings with the landing skids at least touching the landing pad, achieving an average error of 0.19m. It also demonstrates successful recovery and landing when the landing pad is temporarily obscured. △ Less

Submitted 6 March, 2024; originally announced March 2024.

Comments: 7 pages, 6 figures, 2 tables

The Membership Degree Min-Max Localisation Algorithm

Authors: Thomas Hillebrandt, Heiko Will, Marcel Kyas

Abstract: We introduce the Membership Degree Min-Max (MD-Min-Max) localisation algorithm as a precise and simple lateration algorithm for indoor localisation. MD-Min-Max is based on the well-known Min-Max algorithm that computes a bounding box to estimate the position. MD-Min-Max uses a Membership Function (MF) based on an estimated error distribution of the distance measurements to improve the precision of… ▽ More We introduce the Membership Degree Min-Max (MD-Min-Max) localisation algorithm as a precise and simple lateration algorithm for indoor localisation. MD-Min-Max is based on the well-known Min-Max algorithm that computes a bounding box to estimate the position. MD-Min-Max uses a Membership Function (MF) based on an estimated error distribution of the distance measurements to improve the precision of Min-Max. The algorithm has similar complexity to Min-Max and can be used for indoor localisation even on small devices, e.g., in Wireless Sensor Networks (WSNs). To evaluate the performance of the algorithm, we compare it with other improvements of the Min-Max algorithm and maximum likelihood estimators, both in simulations and in a large real-world deployment of a WSN. Results show that MD-Min-Max achieves the best performance in terms of average positioning accuracy while keeping computational cost low compared to the other algorithms. △ Less

Submitted 16 March, 2023; originally announced March 2023.

Comments: This paper was invited and accepted for publication in a special issue of the "Journal of Global Positioning Systems" in 2013. That issue has never appeared. After a decade, we decided to publish the final manuscript

Autonomous Drone Landing: Marked Landing Pads and Solidified Lava Flows

Authors: Joshua Springer, Marcel Kyas

Abstract: Landing is the most challenging and risky aspect of multirotor drone flight, and only simple landing methods exist for autonomous drones. We explore methods for autonomous drone landing in two scenarios. In the first scenario, we examine methods for landing on known landing pads using fiducial markers and a gimbal-mounted monocular camera. This method has potential in drone applications where a dr… ▽ More Landing is the most challenging and risky aspect of multirotor drone flight, and only simple landing methods exist for autonomous drones. We explore methods for autonomous drone landing in two scenarios. In the first scenario, we examine methods for landing on known landing pads using fiducial markers and a gimbal-mounted monocular camera. This method has potential in drone applications where a drone must land more accurately than GPS can provide (e.g.~package delivery in an urban canyon). We expand on previous methods by actuating the drone's camera to track the marker over time, and we address the complexities of pose estimation caused by fiducial marker orientation ambiguity. In the second scenario, and in collaboration with the RAVEN project, we explore methods for landing on solidified lava flows in Iceland, which serves as an analog environment for Mars and provides insight into the effectiveness of drone-rover exploration teams. Our drone uses a depth camera to visualize the terrain, and we are developing methods to analyze the terrain data for viable landing sites in real time with minimal sensors and external infrastructure requirements, so that the solution does not heavily influence the drone's behavior, mission structure, or operational environments. △ Less

Submitted 1 February, 2023; originally announced February 2023.

Comments: 10 pages, 12 figures

Autonomous Drone Landing with Fiducial Markers and a Gimbal-Mounted Camera for Active Tracking

Authors: Joshua Springer, Marcel Kyas

Abstract: Precision landing is a remaining challenge in autonomous drone flight. Fiducial markers provide a computationally cheap way for a drone to locate a landing pad and autonomously execute precision landings. However, most work in this field depends on either rigidly-mounted or downward-facing cameras which restrict the drone's ability to detect the marker. We present a method of autonomous landing th… ▽ More Precision landing is a remaining challenge in autonomous drone flight. Fiducial markers provide a computationally cheap way for a drone to locate a landing pad and autonomously execute precision landings. However, most work in this field depends on either rigidly-mounted or downward-facing cameras which restrict the drone's ability to detect the marker. We present a method of autonomous landing that uses a gimbal-mounted camera to quickly search for the landing pad by simply spinning in place while tilting the camera up and down, and to continually aim the camera at the landing pad during approach and landing. This method demonstrates successful search, tracking, and landing with 4 of 5 tested fiducial systems on a physical drone with no human intervention. Per fiducial system, we present the distributions of the distances from the drone to the center of the landing pad after each successful landing. We also show representative examples of flight trajectories, marker tracking performance, and control outputs for each channel during the landing. Finally, we discuss qualitative strengths and weaknesses underlying each system. △ Less

Submitted 11 November, 2022; v1 submitted 9 June, 2022; originally announced June 2022.

Comments: Update for IRC

Evaluation of Orientation Ambiguity and Detection Rate in April Tag and WhyCode

Authors: Joshua Springer, Marcel Kyas

Abstract: Fiducial systems provide a computationally cheap way for mobile robots to estimate the pose of objects, or their own pose, using just a monocular camera. However, the orientation component of the pose of fiducial markers is unreliable, which can have destructive effects in autonomous drone landing on landing pads marked with fiducial markers. This paper evaluates the April Tag and WhyCode fiducial… ▽ More Fiducial systems provide a computationally cheap way for mobile robots to estimate the pose of objects, or their own pose, using just a monocular camera. However, the orientation component of the pose of fiducial markers is unreliable, which can have destructive effects in autonomous drone landing on landing pads marked with fiducial markers. This paper evaluates the April Tag and WhyCode fiducial systems in terms of orientation ambiguity and detection rate on embedded hardware. We test 2 April Tag variants - 1 default and 1 custom - and 3 Whycode variants - 1 default and 2 custom. We determine that they are suitable for autonomous drone landing applications in terms of detection rate, but may generate erroneous control signals as a result of orientation ambiguity in the pose estimates. △ Less

Submitted 11 November, 2022; v1 submitted 18 March, 2022; originally announced March 2022.

Comments: Update for IRC paper