This paper presents control and coordination algorithms for groups of vehicles. The focus is on autonomous vehicle networks performing distributed sensing tasks where each vehicle plays the role of a mobile tunable sensor. The paper proposes gradient descent algorithms for a class of utility functions which encode optimal coverage and sensing policies. The resulting closed-loop behavior is adaptive, distributed, asynchronous, and verifiably correct.

Coverage control for mobile sensing networks

In the real world, a realistic setting for computer vision or multimedia recognition problems is that we have some classes containing lots of training data and many classes contain a small amount of training data. Therefore, how to use frequent classes to help learning rare classes for which it is harder to collect the training data is an open question. Learning with Shared Information is an emerging topic in machine learning, computer vision and multimedia analysis. There are different level of components that can be shared during concept modeling and machine learning stages, such as sharing generic object parts, sharing attributes, sharing transformations, sharing regularization parameters and sharing training examples, etc. Regarding the specific methods, multi-task learning, transfer learning and deep learning can be seen as using different strategies to share information. These learning with shared information methods are very effective in solving real-world large-scale problems. This special issue aims at gathering the recent advances in learning with shared information methods and their applications in computer vision and multimedia analysis. Both state-of-the-art works, as well as literature reviews, are welcome for submission. Papers addressing interesting real-world computer vision and multimedia applications are especially encouraged. Topics of interest include, but are not limited to:  • Multi-task learning or transfer learning for large-scale computer vision and multimedia analysis • Deep learning for large-scale computer vision and multimedia analysis • Multi-modal approach for large-scale computer vision and multimedia analysis • Different sharing strategies, e.g., sharing generic object parts, sharing attributes, sharing transformations, sharing regularization parameters and sharing training examples, • Real-world computer vision and multimedia applications based on learning with shared information, e.g., event detection, object recognition, object detection, action recognition, human head pose estimation, object tracking, location-based services, semantic indexing. • New datasets and metrics to evaluate the benefit of the proposed sharing ability for the specific computer vision or multimedia problem. • Survey papers regarding the topic of learning with shared information.  Authors who are unsure whether their planned submission is in scope may contact the guest editors prior to the submission deadline with an abstract, in order to receive feedback.

IEEE transactions on pattern analysis and machine intelligence

Morphometric tools for landmark data: geometry and biology

In recent years there have been excellent results in visual-inertial odometry techniques, which aim to compute the incremental motion of the sensor with high accuracy and robustness. However, these approaches lack the capability to close loops and trajectory estimation accumulates drift even if the sensor is continually revisiting the same place. In this letter, we present a novel tightly coupled visual-inertial simultaneous localization and mapping system that is able to close loops and reuse its map to achieve zero-drift localization in already mapped areas. While our approach can be applied to any camera configuration, we address here the most general problem of a monocular camera, with its well-known scale ambiguity. We also propose a novel IMU initialization method, which computes the scale, the gravity direction, the velocity, and gyroscope and accelerometer biases, in a few seconds with high accuracy. We test our system in the 11 sequences of a recent micro-aerial vehicle public dataset achieving a typical scale factor error of 1% and centimeter precision. We compare to the state-of-the-art in visual-inertial odometry in sequences with revisiting, proving the better accuracy of our method due to map reuse and no drift accumulation.

/pdf/visual-inertial-monocular-slam-with-map-reuse-qzzpcqq1e3.pdf

Visual-Inertial Monocular SLAM With Map Reuse

Over the past few years, the application of camera-equipped Unmanned Aerial Vehicles (UAVs) for visually monitoring construction and operation of buildings, bridges, and other types of civil infrastructure systems has exponentially grown. These platforms can frequently survey construction sites, monitor work-in-progress, create documents for safety, and inspect existing structures, particularly for hard-to-reach areas. The purpose of this paper is to provide a concise review of the most recent methods that streamline collection, analysis, visualization, and communication of the visual data captured from these platforms, with and without using Building Information Models (BIM) as a priori information. Specifically, the most relevant works from Civil Engineering, Computer Vision, and Robotics communities are presented and compared in terms of their potential to lead to automatic construction monitoring and civil infrastructure condition assessment.

/pdf/visual-monitoring-of-civil-infrastructure-systems-via-camera-7xo7w3ojtr.pdf

Visual monitoring of civil infrastructure systems via camera-equipped Unmanned Aerial Vehicles (UAVs): a review of related works

The fusion of IMU and RGB-D sensors presents an interesting combination of information to achieve autonomous localization and mapping using robotic platforms such as ground robots and flying vehicles. In this paper, we present a software framework for cooperative localization and mapping while simultaneously using multiple aerial platforms. We employ a monocular visual odometry algorithm to solve the localization task, where the depth data flow associated to the RGB image is used to estimate the scale factor associated with the visual information. The current framework enables autonomous onboard control of each vehicle with cooperative localization and mapping. We present a methodology that provides both a sparse map generated by the monocular SLAM and a multiple resolution dense map generated by the associated depth. The localization algorithm and both 3D mapping algorithms work in parallel improving the system real-time reliability. We present experimental results to show the effectiveness of the proposed approach using two quadrotors platforms.

Cooperative localization and mapping of MAVs using RGB-D sensors

The combination of on-board sensors measurements with different statistical characteristics can be employed in robotics for localization and control, especially in GPS-denied environments. In particular, most aerial vehicles are packaged with low cost sensors, important for aerial robotics, such as camera, a gyroscope, and an accelerometer. In this work, we develop a visual inertial odometry system based on the Unscented Kalman Filter (UKF) acting on the Lie group SE(3), such to obtain an unique, singularity-free representation of a rigid body pose. We model this pose with the Lie group SE(3) and model the noise on the corresponding Lie algebra. Moreover, we extend the concepts used in the standard UKF formulation, such as state uncertainty and modeling, to correctly incorporate elements that do not belong to an Euclidean space such as the Lie group members. In this analysis, we use the parallel transport, which requires us to explicitly consider SE(3) as representing rigid bodies though the use of the affine connection. We present experimental results to show the effectiveness of the proposed approach for state estimation of a quadrotor platform.

Visual inertial odometry for quadrotors on SE(3)

We propose a lightweight embedded system for stabilization and control of Unmanned Aerial Vehicles (UAVs) and particularly Micro Aerial Vehicles (MAVs). The system relies solely on onboard sensors to localize the MAV, which makes it suitable for experiments in GPS-denied environments. The system utilizes predictive controllers to find optimal control actions for the aircraft using only onboard computational resources. To show the practicality of the proposed system, we present several indoor and outdoor experiments with multiple quadrotor helicopters which demonstrate its capability of trajectory tracking and disturbance rejection.

Embedded model predictive control of unmanned micro aerial vehicles

Consumer grade technology seen in cameras and phones has led to the price/performance ratio of sensors and processors falling dramatically over the last decade. In particular, most devices are packaged with a camera, a gyroscope, and an accelerometer, important sensors for aerial robotics. The low mass and small form factor make them particularly well suited for autonomous flight with small flying robots, especially in GPS-denied environments. In this work, we present the first fully autonomous smartphone-based quadrotor. All the computation, sensing and control runs on an off-the-shelf smartphone, with all the software functionality in a smartphone app.We show how quadrotors can be stabilized and controlled to achieve autonomous flight in indoor buildings with application to smart homes, search and rescue, construction and architecture. The work allows any consumer with a smartphone to autonomously drive a quadrotor robot platform, even without GPS, by downloading an app, and concurrently build 3-D maps.

Smartphones power flying robots

A new vision-based obstacle avoidance technique for indoor navigation of Micro Aerial Vehicles (MAVs) is presented in this paper. The vehicle trajectory is modified according to the obstacles detected through the Depth Map of the surrounding environment, which is computed online using the Optical Flow provided by a single onboard omnidirectional camera. An existing closed-form solution for the absolute-scale velocity estimation based on visual correspondences and inertial measurements is generalized and here employed for the Depth Map estimation. Moreover, a dynamic region-of-interest for image features extraction and a self-limitation control for the navigation velocity are proposed to improve safety in view of the estimated vehicle velocity. The proposed solutions are validated by means of simulations.

/pdf/mav-indoor-navigation-based-on-a-closed-form-solution-for-ls30efyi21.pdf

Giuseppe Loianno

Papers

Cooperative localization and mapping of MAVs using RGB-D sensors

Visual inertial odometry for quadrotors on SE(3)

Embedded model predictive control of unmanned micro aerial vehicles

Smartphones power flying robots

MAV indoor navigation based on a closed-form solution for absolute scale velocity estimation using Optical Flow and inertial data