Figure 14 - available via license: Creative Commons Attribution 4.0 International
Content may be subject to copyright.
Yaw-axis gimbal hardware setup developed by the authors. A monocular camera has been mounted on a Dynamixal stepper motor, which is controlled by an Arduino Mega 2560 controller. The controller is used as a slave ROS process in localization application. Housing is in a 3D printed retrofit.
Source publication
In this paper, we analyzed the accuracy and precision of AprilTag as a visual fiducial marker in detail. We have analyzed error propagation along two horizontal axes along with the effect of angular rotation about the vertical axis. We have identified that the angular rotation of the camera (yaw angle) about its vertical axis is the primary source...
Citations
... Two processing techniques were proposed by [15], including a method that can filter out very inaccurate detections resulting from partial border occlusion, and a new highly accurate method for edge refinement. Based on observations, ref. [16] proposed improvements include trigonometric yaw angle correction for camera alignment, real-time tag center tracking via a custom yaw-axis gimbal, and a novel poseindexed probabilistic sensor error model for AprilTag utilizing Gaussian Processes and validated by particle filter tracking. In the process of vehicle movement, ref. [17] found that the marker detection error ranged from 0.028 m to 0.13 m. ...
In environments where Global Navigation Satellite System (GNSS) signals are unavailable, our proposed multi-tag fusion localization method offers a robust solution for the precise positioning of vehicles or robots. During our research, we observed variations in the positioning information estimated from tags located at different positions within the same frame. Our goal was to extract reliable positioning information from this noisy data. By constructing geometric constraints, our method introduces an outlier factor to quantify the differences between tags. After effectively eliminating outliers, we enhanced the Kalman filter framework to accommodate the fusion of data from two or more tags, with the outlier factor dynamically adjusting the observation noise during the fusion process. The experimental results demonstrate that, even under the influence of motion and obstacles, our method maintains position errors within a 3 cm range and orientation errors within 3°. This indicates that our method possesses high positioning accuracy and stability.
... computational simplicity. However, its use as a localization system may result in erroneous localization due to factors such as viewing angle, distance, and camera rotation [25]. Furthermore, in this investigation, the UGV's state is ascertained utilizing standard landing helipads through the implementation of YOLOv5, a rapid and highly accurate object detection algorithm using a deep learning model developed in PyTorch. ...
... April tag is advantageous for its low cost and computational simplicity. However, its use as a localization system may result in erroneous localization due to factors such as viewing angle, distance, and camera rotation [25]. ...
This paper presents a vision-based adaptive tracking and landing method for multirotor Unmanned Aerial Vehicles (UAVs), designed for safe recovery amid propulsion system failures that reduce maneuverability and responsiveness. The method addresses challenges posed by external disturbances such as wind and agile target movements, specifically, by considering maneuverability and control limitations caused by propulsion system failures. Building on our previous research in actuator fault detection and tolerance, our approach employs a modified adaptive pure pursuit guidance technique with an extra adaptation parameter to account for reduced maneuverability, thus ensuring safe tracking of moving objects. Additionally, we present an adaptive landing strategy that adapts to tracking deviations and minimizes off-target landings caused by lateral tracking errors and delayed responses, using a lateral offset-dependent vertical velocity control. Our system employs vision-based tag detection to ascertain the position of the Unmanned Ground Vehicle (UGV) in relation to the UAV. We implemented this system in a mid-mission emergency landing scenario, which includes actuator health monitoring of emergency landings. Extensive testing and simulations demonstrate the effectiveness of our approach, significantly advancing the development of safe tracking and emergency landing methods for UAVs with compromised control authority due to actuator failures.
... To estimate the camera pose, researchers concentrate on two main methods, namely the marked and the unmarked method. The marked method utilizes a known pattern as ground truth, in which the camera detects the patterns knowing their world positions [12], [13], [14]. This method is simple and highly accurate as it is purposefully designed in a high-precision setup; however, it requires extensive labor and is not wellsuited for real world robot applications due to frequent recalibration. ...
Calibration is the first and foremost step in dealing with sensor displacement errors that can appear during extended operation and off-time periods to enable robot object manipulation with precision. In this paper, we present a novel multiplanar self-calibration between the camera system and the robot's end-effector for 3D object manipulation. Our approach first takes the robot end-effector as ground truth to calibrate the camera's position and orientation while the robot arm moves the object in multiple planes in 3D space, and a 2D state-of-the-art vision detector identifies the object's center in the image coordinates system. The transformation between world coordinates and image coordinates is then computed using 2D pixels from the detector and 3D known points obtained by robot kinematics. Next, an integrated stereo-vision system estimates the distance between the camera and the object, resulting in 3D object localization. We test our proposed method on the Baxter robot with two 7-DOF arms and a 2D detector that can run in real time on an onboard GPU. After self-calibrating, our robot can localize objects in 3D using an RGB camera and depth image. The source code is available at https://github.com/tuantdang/calib_cobot.
... The recognition distance of Apriltag is related to many factors [11]: ...
... The extracted pattern is then validated against a library of possible code schemes. In the last step, the six-degrees-of-freedom (DoF) pose of the marker in the camera reference frame is estimated by computing the homographic transform [12,13]. Like other solutions for mobile robot localisation, planar fiducial markers have their own shortcomings: ...
... Other authors proposed larger markers in combination with alternative patterns such as random dots proposed by Uchiyama et al. [34] or the uniform marker field proposed by Szentandrási et al. [17]. Another group of articles recommends the use of 3D markers [5,13,35]; however, the usage of these markers might be inconvenient, e.g., larger spatial requirements or a geometric model has to be very accurate [33]. ...
... The function f (m c , m g ) represents a transformation of the position and orientation of the camera and consequently of the robot itself from the camera coordinate system to the global coordinate system. The known pose of the marker in the global coordinate system m g and in the camera coordinate system m c needs to be combined using the geometric transformations in Equations (11)- (13) to obtain the pose of the camera in the global coordinate system c g . ...
Planar fiducial markers are commonly used to estimate a pose of a camera relative to the marker. This information can be combined with other sensor data to provide a global or local position estimate of the system in the environment using a state estimator such as the Kalman filter. To achieve accurate estimates, the observation noise covariance matrix must be properly configured to reflect the sensor output’s characteristics. However, the observation noise of the pose obtained from planar fiducial markers varies across the measurement range and this fact needs to be taken into account during the sensor fusion to provide a reliable estimate. In this work, we present experimental measurements of the fiducial markers in real and simulation scenarios for 2D pose estimation. Based on these measurements, we propose analytical functions that approximate the variances of pose estimates. We demonstrate the effectiveness of our approach in a 2D robot localisation experiment, where we present a method for estimating covariance model parameters based on user measurements and a technique for fusing pose estimates from multiple markers.
... • Visual guided with AprilTags [4]: One example of this solution is the ACE™ -Autonomous Control Engine system commercialized by Planck AeroSystems 1 for autonomous landing on moving vessel. It consists of a large AprilTag mounted on top of the moving platform which is tracked by a camera onboard the UAV. ...
... • ROS TF Transform allows ROS to display multiple coordinate frames with respect to time. • April Tag 3 is a comprehensive open-source visual fiducial marker detection algorithm [37]. The function of the tags is to provide a precise 3D position that can be referenced against the point cloud. ...
Localization is a keystone for a robot to work within its environment and with other robots. There have been many methods used to solve this problem. This paper deals with the use of beacon-based localization to answer the research question: Can ultra-wideband technology be used to effectively localize a robot with sensor fusion? This paper has developed an innovative solution for creating a sensor fusion platform that uses ultra-wideband communication as a localization method to allow an environment to be perceived and inspected in three dimensions from multiple perspectives simultaneously. A series of contributions have been presented, supported by an in-depth literature review regarding topics in this field of knowledge. The proposed method was then designed, built, and tested successfully in two different environments exceeding its required tolerances. The result of the testing and the ideas formulated throughout the paper were discussed and future work outlined on how to build upon this work in potential academic papers and projects.
... Unlike Quick Response (QR) codes, these markers contain a small information payload, and thus, they can be quickly detected even in different lighting conditions. However, to estimate the pose of the MAV using the visual fiducial markers, its accuracy and robustness depend on many factors, such as the camera calibration parameters, image resolution, distance to the markers, viewing angles, and processing speed or vehicle motion [29]. Therefore, to achieve accurate indoor localization, many markers need to be placed as close as possible or in such a way that at least one marker is fully visible in the camera field of view along all possible flight routes. ...
Micro air vehicles (MAVs) are ideal for indoor precision farming due to their agility and minimal impact on surrounding objects. To achieve autonomous operations, these vehicles hardly rely on the commonly used Global Positioning System for indoor navigation and monitor plants with severely limited payloads. This paper proposes an autonomous MAV system that incorporates an ultra-wideband (UWB) system and a deep learning method to perform indoor navigation and localize unhealthy plants. The position of the MAV is estimated using an extended Kalman filter based on the trilateration results of the ranging measurements obtained from the UWB system. A multiple bounding box prediction strategy is used to analyze all the leaves of the plants and quickly identify their health conditions. Several deep learning models were selected and trained to detect unhealthy plants using field condition images. The model that achieved the desired combination of the accuracy and efficiency was fine-tuned with various image resolutions and sizes of negative samples to further improve its detection accuracy. Several real-world flight tests were performed successfully using the UWB system, and the plants were classified correctly with the selected model.
... The decoding of tags is by calculating the relative coordinates of the tags of each bit field, using homography to convert them into image coordinates, and then thresholding the resulting pixels. In order to reduce the influence of light on the recognition effect, the intensity "black" pixel of the spatial variation model and the intensity "white" model of the second model were established in Apriltag [4]. The expression is shown in formula (1). ...
In the research and development of target tracking, vision-based target tracking technology still has the problems of low accuracy and high cost. This paper designs a tracking system that uses a four-wheel differential mobile chassis as a carrier and uses the 36H11 series tags in Apriltag as a moving target. STM32F103 single-chip microcomputer is used as the core of motion control, and the classic PID control algorithm is used to adjust the wheel speed to achieve target tracking. STM32F765 single-chip microcomputer is used as the image processor of the OV7725 camera to solve the label information. The experimental results show that the Apriltag tag target can be better tracked when the PID parameters are adjusted properly. It can achieve near real-time tracking effect when the tag moving speed is slow, and it can move to the specified position quickly when it is far away from the target.
... Of the provided options, the survey reflectors gave the most accurate and consistent transforms for our systems. Separate analyses concluded that the position variance of Leica prisms is on the order of millimeters (Lackner and Lienhart, 2016), consistent with the centering accuracy specified by Leica (Leica Geosystems, 2020), whereas Apriltag variance is on the order of centimeters (Abbas et al., 2019). Each system is equipped with a set of three survey prisms, and both the robot and gate transforms are estimated in the survey station's frame using Horn's absolute orientation method (Horn, 1987). ...
... Pose estimates generate from Apriltags had high variance since the robots' front facing camera could only localize all three Apriltags far from any tag and the camera viewed the left and right Apriltags with yaw from the tags center. Both a sensor's distance (Wang and Olson, 2016) and yaw angle (Abbas et al., 2019) are known to decrease estimated pose accuracy, and even these relatively minor effects led to significant pose estimate errors. For the first Tunnel Circuit deployment ICP alignment conducted after the fact estimated the difference between the first robot's map and the third robot's map in the world frame was 0.0527 deg of roll, 3.9215 deg of pitch, and −0.4105 deg of yaw in the worst case. ...
Artificial intelligence has undergone immense growth and maturation in recent years, though autonomous systems have traditionally struggled when fielded in diverse and previously unknown environments. DARPA is seeking to change that with the Subterranean Challenge, by providing roboticists the opportunity to support civilian and military first responders in complex and high-risk underground scenarios. The subterranean domain presents a handful of challenges, such as limited communication, diverse topology and terrain, and degraded sensing. Team MARBLE proposes a solution for autonomous exploration of unknown subterranean environments in which coordinated agents search for artifacts of interest. The team presents two navigation algorithms in the form of a metric-topological graph-based planner and a continuous frontier-based planner. To facilitate multi-agent coordination, agents share and merge new map information and candidate goal points. Agents deploy communication beacons at different points in the environment, extending the range at which maps and other information can be shared. Onboard autonomy reduces the load on human supervisors, allowing agents to detect and localize artifacts and explore autonomously outside established communication networks. Given the scale, complexity, and tempo of this challenge, a range of lessons was learned, most importantly, that frequent and comprehensive field testing in representative environments is key to rapidly refining system performance.
