Hriday Bavle

Natural user interfaces for human-drone multi-modal interaction

Personal drones are becoming part of every day life. To fully integrate them into society, it is crucial to design safe and intuitive ways to interact with these aerial systems. The recent advances on User-Centered Design (UCD) applied to Natural User Interfaces (NUIs) intend to make use of human innate features, such as speech, gestures and vision to interact with technology in the way humans would with one another. In this paper, a Graphical User Interface (GUI) and several NUI methods are studied and implemented, along with computer vision techniques, in a single software framework for aerial robotics called Aerostack which allows for intuitive and natural human-quadrotor interaction in indoor GPS-denied environments. These strategies include speech, body position, hand gesture and visual marker interactions used to directly command tasks to the drone. The NUIs presented are based on devices like the Leap Motion Controller, microphones and small size monocular on-board cameras which are unnoticeable to the user. Thanks to this UCD perspective, the users can choose the most intuitive and effective type of interaction for their application. Additionally, the strategies proposed allow for multi-modal interaction between multiple users and the drone by being able to integrate several of these interfaces in one single application as is shown in various real flight experiments performed with non-expert users.

AEROSTACK: An architecture and open-source software framework for aerial robotics

To simplify the usage of the Unmanned Aerial Systems (UAS), extending their use to a great number of applications, fully autonomous operation is needed. There are many open-source architecture frameworks for UAS that claim the autonomous operation of UAS, but they still have two main open issues: (1) level of autonomy, being in most of the cases limited and (2) versatility, being most of them designed specifically for some applications or aerial platforms. As a response to these needs and issues, this paper presents Aerostack, a system architecture and open-source multi-purpose software framework for autonomous multi-UAS operation. To provide higher degrees of autonomy, Aerostacks system architecture integrates state of the art concepts of intelligent, cognitive and social robotics, based on five layers: reactive, executive, deliberative, reflective, and social. To be a highly versatile practical solution, Aerostacks open-source software framework includes the main components to execute the architecture for fully autonomous missions of swarms of UAS; a collection of ready-to-use and flight proven modular components that can be reused by the users and developers; and compatibility with five well known aerial platforms, as well as a high number of sensors. Aerostack has been validated during three years by its successful use on many research projects, international competitions and exhibitions. To corroborate this fact, this paper also presents Aerostack carrying out a fictional fully autonomous indoors search and rescue mission.

A flexible and dynamic mission planning architecture for UAV swarm coordination

In this paper a scalable and flexible Architecture for real-time mission planning and dynamic agent-to-task assignment for a swarm of Unmanned Aerial Vehicles (UAV) is presented. The proposed mission planning architecture consists of a Global Mission Planner (GMP) which is responsible of assigning and monitoring different high-level missions through an Agent Mission Planner (AMP), which is in charge of providing and monitoring each task of the mission to each UAV in the swarm. The objective of the proposed architecture is to carry out high-level missions such as autonomous multi-agent exploration, automatic target detection and recognition, search and rescue, and other different missions with the ability of dynamically re-adapt the mission in real-time. The proposed architecture has been evaluated in simulation and real indoor flights demonstrating its robustness in different scenarios and its flexibility for real-time mission re-planning and dynamic agent-to-task assignment.

A Multi-Layered Component-Based Approach for the Development of Aerial Robotic Systems: The Aerostack Framework

To achieve fully autonomous operation for Unmanned Aerial Systems (UAS) it is necessary to integrate multiple and heterogeneous technical solutions (e.g., control-based methods, computer vision methods, automated planning, coordination algorithms, etc.). The combination of such methods in an operational system is a technical challenge that requires efficient architectural solutions. In a robotic engineering context, where productivity is important, it is also important to minimize the effort for the development of new systems. As a response to these needs, this paper presents Aerostack, an open-source software framework for the development of aerial robotic systems. This framework facilitates the creation of UAS by providing a set of reusable components specialized in functional tasks of aerial robotics (trajectory planning, self localization, etc.) together with an integration method in a multi-layered cognitive architecture based on five layers: reactive, executive, deliberative, reflective and social. Compared to other software frameworks for UAS, Aerostack can provide higher degrees of autonomy and it is more versatile to be applied to different types of hardware (aerial platforms and sensors) and different types of missions (e.g. multi robot swarm systems). Aerostack has been validated during four years (since February 2013) by its successful use on many research projects, international competitions and public exhibitions. As a representative example of system development, this paper also presents how Aerostack was used to develop a system for a (fictional) fully autonomous indoors search and rescue mission.

A fully-autonomous aerial robotic solution for the 2016 International Micro Air Vehicle competition

In this paper, a fully-autonomous quadrotor aerial robot for solving the different missions proposed in the 2016 International Micro Air Vehicle (IMAV) Indoor Competition is presented. The missions proposed in the IMAV 2016 competition involve the execution of high-level missions such as entering and exiting a building, exploring an unknown indoor environment, recognizing and interacting with objects, landing autonomously on a moving platform, etc. For solving the aforementioned missions, a fully-autonomous quadrotor aerial robot has been designed, based on a complete hardware configuration and a versatile software architecture, which allows the aerial robot to complete all the missions in a fully autonomous and consecutive manner. A thorough evaluation of the proposed system has been carried out in both simulated flights, using the Gazebo simulator in combination with PX4 Software-In-The-Loop, and real flights, demonstrating the appropriate capabilities of the proposed system for performing high-level missions and its flexibility for being adapted to a wide variety of applications.

A flight altitude estimator for multirotor UAVs in dynamic and unstructured indoor environments

A reliable estimation of the flight altitude in dynamic and unstructured indoor environments is an unsolved problem. Standalone available sensors, such as distance sensors, barometers and accelerometers, have multiple limitations in presence of non-flat ground surfaces, or in cluttered areas. To overcome these sensor limitations, maximizing their individual performance, this paper presents a modular EKF- based multi-sensor fusion approach for accurate vertical localization of multirotor UAVs in dynamic and unstructured indoor environments. The state estimator allows to combine the information provided by a variable number and type of sensors, including IMU, barometer and distance sensors, with the capabilities of sensor auto calibration and bias estimation, as well as a flexible configuration of the prediction and update stages. Several autonomous indoors real flights in unstructured environments have been conducted in order to validate our proposed state estimator, enabling the UAV to maintain the desired flight altitude when navigating over wide range of obstacles. Furthermore, it has been successfully used in IMAV 2016 competition. The presented work has been made publicly available to the scientific community as an open source software within the Aerostack1 framework.

Towards fully autonomous landing on moving platforms for rotary Unmanned Aerial Vehicles

Fully autonomous landing on moving platforms poses a problem of importance for Unmanned Aerial Vehicles (UAVs). Current approaches are usually based on tracking and following the moving platform by means of several techniques, which frequently lack performance in real applications. The aim of this paper is to prove a simple landing strategy is able to provide practical results. The presented approach is based on three stages: estimation, prediction and fast landing. As a preliminary phase, the problem is solved for a particular case of the IMAV 2016 competition. Subsequently, it is extended to a more generic and versatile approach. A thorough evaluation has been conducted with simulated and real flight experiments. Simulations have been performed utilizing Gazebo 6 and PX4 Software-In-The-Loop (SITL) and real flight experiments have been conducted with a custom quadrotor and a moving platform in an indoor environment.

A Deep Reinforcement Learning Strategy for UAV Autonomous Landing on a Moving Platform

The use of multi-rotor UAVs in industrial and civil applications has been extensively encouraged by the rapid innovation in all the technologies involved. In particular, deep learning techniques for motion control have recently taken a major qualitative step, since the successful application of Deep Q-Learning to the continuous action domain in Atari-like games. Based on these ideas, Deep Deterministic Policy Gradients (DDPG) algorithm was able to provide outstanding results with continuous state and action domains, which are a requirement in most of the robotics-related tasks. In this context, the research community is lacking the integration of realistic simulation systems with the reinforcement learning paradigm, enabling the application of deep reinforcement learning algorithms to the robotics field. In this paper, a versatile Gazebo-based reinforcement learning framework has been designed and validated with a continuous UAV landing task. The UAV landing maneuver on a moving platform has been solved by means of the novel DDPG algorithm, which has been integrated in our reinforcement learning framework. Several experiments have been performed in a wide variety of conditions for both simulated and real flights, demonstrating the generality of the approach. As an indirect result, a powerful work flow for robotics has been validated, where robots can learn in simulation and perform properly in real operation environments. To the best of the authors knowledge, this is the first work that addresses the continuous UAV landing maneuver on a moving platform by means of a state-of-the-art deep reinforcement learning algorithm, trained in simulation and tested in real flights.

Attitude estimation using horizon detection in thermal images

The lack of redundant attitude sensors represents a considerable yet common vulnerability in many low-cost unmanned aerial vehicles. In addition to the use of attitude sensors, exploiting the horizon as a visual reference for attitude control is part of human pilots’ training. For this reason, and given the desirable properties of image sensors, quite a lot of research has been conducted proposing the use of vision sensors for horizon detection in order to obtain redundant attitude estimation onboard unmanned aerial vehicles. However, atmospheric and illumination conditions may hinder the operability of visible light image sensors, or even make their use impractical, such as during the night. Thermal infrared image sensors have a much wider range of operation conditions and their price has greatly decreased during the last years, becoming an alternative to visible spectrum sensors in certain operation scenarios. In this paper, two attitude estimation methods are proposed. The first method consists of a novel approach to estimate the line that best fits the horizon in a thermal image. The resulting line is then used to estimate the pitch and roll angles using an infinite horizon line model. The second method uses deep learning to predict attitude angles using raw pixel intensities from a thermal image. For this, a novel Convolutional Neural Network architecture has been trained using measurements from an inertial navigation system. Both methods presented are proven to be valid for redundant attitude estimation, providing RMS errors below 1.7° and running at up to 48 Hz, depending on the chosen method, the input image resolution and the available computational capabilities.