Showing papers presented at "International Conference on Control, Automation, Robotics and Vision in 2016"

Autonomous navigation of UAV by using real-time model-based reinforcement learning

Abstract: Autonomous navigation in an unknown or uncertain environment is one of the challenging tasks for unmanned aerial vehicles (UAVs). In order to address this challenge, it is necessary to have sophisticated high level control methods that can learn and adapt themselves to changing conditions. One of the most promising frameworks for such a purpose is reinforcement learning. In this paper, a novel model-based reinforcement learning algorithm, TEXPLORE, is developed as a high level control method for autonomous navigation of UAVs. The developed approach has been extensively tested with a quadcopter UAV in ROS-Gazebo environment. The experimental results show that our method is able to learn an efficient trajectory in a few iterations and perform actions in real-time. Moreover, we show that our approach significantly outperforms Q-learning based method. To the best of our knowledge, this is the first time that TEXPLORE has been developed to achieve autonomous navigation of UAVs.

Adaptive consensus disturbance rejection for multi-agent systems on directed graphs

Abstract: In this paper, the adaptive consensus disturbance rejection problem is considered for the liner multi-agent systems under directed graphs. Based on the relative state information of the neighboring agents, the consensus protocols, including a state observer and a disturbance observer, are designed to guarantee that the consensus error goes to zero with the complete disturbance rejection. Furthermore, the state observer is designed in a fully distributed fashion with adaptive coupling gain, which has the advantage that the consensus protocol design is independent of the Laplacian matrix associated with the communication network. Finally, an example is given to verify the effectiveness of the theoretical results.

Computer vision-based method for concrete crack detection

Abstract: This paper presents a computer vision-based method to automatically detect concrete cracks. We focus on images containing the concrete: background and crack, where the background is the major mode of the gray-scale histogram. Therefore, we address the detection problem of potential concrete cracks by dealing with histogram thresholding to extract regions of interests from the background. We first employ line emphasis and moving average filters to remove noise from concrete surface images obtained from an inspection robot. The developed algorithm is then applied for automatic detection of significant peaks from the gray-scale histogram of the smoothed image. The biggest peak and its corresponding valley(s) are consequently identified to calculate the threshold value for image binarization. The effectiveness of our proposed method was successfully evaluated on various test images, where cracks could be identified without the requirement of some heuristic reasoning.

An image matching system for autonomous UAV navigation based on neural network

Abstract: This paper proposes an image matching system using aerial images, captured in flight time, and aerial geo-referenced images to estimate the Unmanned Aerial Vehicle (UAV) position in a situation of Global Navigation Satellite System (GNSS) failure. The image matching system is based on edge detection in the aerial and geo-referenced image and posterior automatic image registration of these edge-images (position estimation of UAV). The edge detection process is performed by an Artificial Neural Network (ANN), with an optimal architecture. A comparison with Sobel and Canny edge extraction filters is also provided. The automatic image registration is obtained by a cross-correlation process. The ANN optimal architecture is set by the Multiple Particle Collision Algorithm (MPCA). The image matching system was implemented in a low cost/consumption portable computer. The image matching system has been tested on real flight-test data and encouraging results have been obtained. Results using real flight-test data will be presented.

Feature-based panoramic image stitching

Abstract: In this paper, a panoramic image stitching algorithm is presented. The aim of image stitching is to detect several images of the same scene and merge them to create a larger image. This is achieved by first detecting the overlapping area of the acquired images, and then aligning and blending the seams of the images automatically to create a seamless panoramic image. The experimental testing into the size of the overlapping area, the use of different focal lengths, the use of tilting and panning, and the number of input images were carried out. The results demonstrated that our algorithm is able to stitch correctly with a minimum of 5% overlapping area, of images acquired by a wide angle lens, by a horizontal and vertical rotation, and also of a multitude of input images acquired at unfixed axis.

View planning for 3D shape reconstruction of buildings with unmanned aerial vehicles

Abstract: This paper presents a novel view planning method to generate suitable viewpoints for the reconstruction of the 3D shape of buildings, based on publicly available 2D map data. The proposed method first makes use of 2D map data, along with estimated height information, to generate a rough 3D model of the target building. Randomized sampling procedures are then employed to generate a set of initial candidate viewpoints for the reconstruction process. The most suitable viewpoints are selected from the candidate viewpoint set by first formulating a modified Set Covering Problem (SCP) which considers image registration constraints, as well as uncertainties present in the rough 3D model. A neighborhood greedy search algorithm is proposed to solve this SCP problem and select a series of individual viewpoints deemed most suitable for the 3D reconstruction task. The paper concludes with both computational and real-world field tests to demonstrate the overall effectiveness of the proposed method.

Emergent formations of a Lagrangian swarm of unmanned ground vehicles

Abstract: In this paper, we present a Lagrangian swarm model that shows emergent formations of multiple autonomous unmanned ground vehicles (UGVs). The Direct Method of Lyapunov is used to construct the instantaneous velocity of each individual in the swarm. The velocity controllers ensure the cohesion and therefore the stability of the swarm. Via computer simulations, we illustrate self-organizations such as concentric circular and concentric elliptic formations.

Online trajectory generation using clothoid segments

Abstract: In this paper, a trajectory generation algorithm within an integrated flight guidance and control system is presented. The approach uses clothoids to deal with the problem of curvature steps during the transition phase between straight line and arc flight. The algorithm is designed in interaction with a trajectory controller for 2nd order error dynamics, which hence requires more trajectory information than a conventional approach. The presented test results were obtained via in-flight tests with an experimental CS 23 aircraft.

Automatic flight path control of an experimental DA42 general aviation aircraft

Abstract: An automatic flight path controller, as part of a modular automatic flight guidance and control system, is presented, along with initial flight test results using a DA42 M-NG flying testbed. The basic principle for the flight path control is a reference model based dynamic inversion of the kinematic equations of motions, with pseudo-control hedging to account for inner loop dynamics and plant response deficits. The kinematic frame flight path commands are transformed into body-frame commands executed by inner loop and autothrust controllers for transverse and linear force control. Initial flight test results are presented, demonstrating the feasibility of the path control approach, with good tracking and disturbance performance already during early flight testing.

RRT-based 3D path planning for formation landing of quadrotor UAVs

Abstract: This paper discusses the formation landing problem of quadrotor UAVs, which is considered as a UAV leader-follower problem, avoiding static obstacles. Rapidly-exploring random tree algorithm is used to generate the path for the leader UAV firstly. In particular, specifics of tree-grow including nodes selection, parent node connection, feasible and optimal path generation are explained. Given the leader UAV position, path finding for the follower UAV is conducted to avoid both static obstacles and the leader quadrotor. Based on the intensive simulations, which are conducted in ROS-Gazebo environment, the proposed framework is considered to be applicable in real-time formation landing of quadrotor UAVs.

Distributed estimation for the unknown orientation of the local reference frames in N-dimensional space

Abstract: In this paper, we propose a novel distributed orientation estimation method of rigid bodies in n-dimensional space using only relative orientation information. For the orientation estimation, n auxiliary variables for each agent are required. A rotation matrix which identifies orientation of local frame with respect to the common frame is obtained by transforming auxiliary variables with the Gram-Schmidt procedure. Since the auxiliary variables are defined on vector space, a consensus-based control law for auxiliary variables achieves a global convergence. Although there exist initial values of auxiliary variables such that auxiliary variables converge to undesired points, we show that the solution of proposed algorithm can be obtained for almost all initial values.

Low speed vehicle localization using WiFi fingerprinting

Abstract: Recently, the problem of fully autonomous navigation of vehicle has gained major interest from research institutes and private companies. In general, these researches rely on GPS in fusion with other sensors to track vehicle in outdoor environment. However, as indoor environment such as car park is also an important scenario for vehicle navigation, the lack of GPS poses a serious problem. This study presents an approach to use WiFi Fingerprinting as a replacement for GPS information in order to allow seamlessly transition of localization architecture from outdoor to indoor environment. Often, movement speed of vehicle in indoor environment is low (10–12km/h) in comparison to outdoor scene but still surpasses human walking speed (3–5km/h, which is usually maximum movement speed for effective WiFi localization). This paper proposes an ensemble classification method together with a motion model in order to deal with the above issue. Experiments show that proposed method is capable of imitating GPS behavior on vehicle tracking.

A probabilistic cellular automata ant memory model for a swarm of foraging robots

Abstract: Foraging is one of the most popular tasks for multi-robot systems and it can be considered a metaphor for a broad class of problems. The complexity of the problem increases proportionally with the number of agents, due to the fact that all robots must act in cooperation to complete the task. The proposed model employs a combination of different natural swarm behavior techniques to control a team of robots. It was named robot probabilistic cellular automata ant memory (RPCAAM). Our inspiration came from the possibility to mimic the cognitive behaviors of foraging ants with those of pedestrians in a building evacuation. The proposed probabilistic homing can avoid inertial behavior near the nests, which improves team performance. Furthermore, some investigations into robot-robot and robot-obstacles conflicts were improved to make the model more adequate to real-world applications. The probabilistic model was contrasted with deterministic homing. Moreover, the proposed method was implemented in a robotics simulation environment called Webots. Simulation results indicate that foraging tasks could be implemented with low complexity in low-cost architectures.

A review on perception-driven obstacle-aided locomotion for snake robots

Abstract: Biological snakes can gracefully traverse a wide range of different and complex environments. Snake robots that can mimic this behaviour could be fitted with sensors and also transport tools to hazardous or confined areas that other robots and humans are unable to access. To carry out such tasks, snake robots must have a high degree of awareness of their surroundings (i.e. perception-driven locomotion) and be capable of efficient obstacle exploitation (i.e. obstacle-aided locomotion) to gain propulsion. These aspects are important to realise the large variety of possible snake robot applications in real-life operations such as fire-fighting, industrial inspection, search-and-rescue and more. In this paper, an elaborate review and discussion of the state-of-the-art, challenges and possibilities of perception-driven obstacle-aided locomotion for snake robots is presented for the first time. Pertinent to snake robots, we focus on current strategies for obstacle avoidance, obstacle accommodation, and obstacle-aided locomotion. Moreover, we put obstacle-aided locomotion into the context of perception and mapping. To this end, we present an overview of relevant key technologies and methods within environment perception, mapping and representation that constitute important aspects of perception-driven obstacle-aided locomotion.

Flightplan flight tests of an experimental DA42 general aviation aircraft

Abstract: The recent emergence of unmanned aerial vehicles asked for both well-performing auto-flight systems and fly-by-wire architectures. Towards this end, the trajectory control module of an integrated auto-flight control system is introduced in this paper, which utilizes nonlinear second-order error dynamics of the position error and uses nonlinear dynamic inversion as a control methodology. Flight test results of a flightplan mission conducted on the institute's general aviation aircraft — a DA42 augmented with experimental fly-by-wire — are presented and the controller's performance is evaluated.

Automated license plate detection using a support vector machine

Abstract: This paper proposes a new method of detecting license plates in images of vehicles where the license plate is shown, and reports the detection results when this method was applied to detection of license plates on vehicles in Japan. This license plate detection process detects only the edge vertical components, and the candidate license plates are narrowed down using the contours obtained by dilation and erosion processing and region fill processing. A SVM (Support Vector Machine) based on negative and positive examples is used to determine whether or not a candidate area is a license plate, and finally the position of the license plate is identified. This study examined how the license plate detection results in license plate and non-license plate images were affected by differences in aspect ratios, differences in brightness between the vehicle body and license plate, and the number of positive and negative examples used for learning. The effectiveness of this method was confirmed to yield a license plate detection rate of approximately 90%.

An instrumentation system for smart monitoring of surface temperature

Abstract: Reliable sensing is a crucial factor for assessing the structural health conditions of civil infrastructure such as sewerage networks, which are susceptible to hydrogen sulfide induced concrete corrosion. In this context, this paper reports the work on the development and characterization of an instrumentation system using an infrared radiometer for monitoring surface temperature variations of the concrete through non-contact measurements. The surface temperature measurements are gathered by positioning the sensor at different distance and angles from the surface of interest. The effects of ambient lighting conditions during measurements are investigated. Furthermore, the sensing performance of the sensor is evaluated by performing statistical error analysis, and the efficacy of a custom-made signal processing board is tested by comparing the electrical signal with reference measures.

Designing a system automation for a novel UAV demonstrator

Abstract: The SAGITTA Research Demonstrator is a flying-wing UAV testbed. Its digital flight control system, which is being developed by the Institute of Flight System Dynamics of the Technical University Munich, consists of cascaded control loops. In a UAV operation scenario, different control loops need to be engaged at different times or connected in different ways to fulfill a given mission. This has to be done with respect to commands from the flight operator or even automatically based on sensor information or data link availability. The task is handled by a system automation module which is part of the flight control system software. Its structure and implementation are introduced and described in this paper.

Sensor fusion of depth camera and ultrasound data for obstacle detection and robot navigation

Abstract: Depth cameras have gained much popularity in robotics in recent years. The Microsoft Kinect camera enables a mobile robot to do essential tasks like localization and navigation. Unfortunately, such structured light cameras also suffer from limitations. Exposing them to direct sunlight renders them blind, and transparent objects like glass windows can not be detected. This is a problem for the task of obstacle detection, where false negative measurements must be avoided. At the same time, ultrasound sensors have been studied by the robotic research community for decades. While they have lost attention with the advent of laser scanners and cameras, they remain successful for special applications due to their robustness and simplicity. In this paper we argue that depth cameras and ultrasound sensors extend each other very well. Ultrasound sensors are able to correct the problems inherent to camera-based sensors. We present a sensor fusion algorithm that merges depth camera data and ultrasound measurements using an occupancy grid approach. We validated the algorithm using obstacles in multiple scenarios.

A vehicular collision warning algorithm based on the time-to-collision estimation under connected environment

Abstract: Considering the traffic safety in the scenario of arterial road with on-ramp, this study proposes a time-to-collision (TTC) based vehicular collision warning algorithm under connected environment. In particular, the information of vehicles of interest, i.e., position, traveling direction and velocity, is assumed to be collected by the roadside device via the vehicle-to-infrastructure (V2I) communications. Then, the TTC of a pair of vehicles in arterial road and on-ramp is estimated based on the position, traveling direction and velocity difference of that pair of vehicles. Consequently, the TTC warning messages can be disseminated to vehicles within the communication range of the roadside device, so as to reduce the risk of collision. The proposed algorithm can be used in the cooperative vehicle infrastructure systems (CVIS) to improve the traffic safety.

Designing a safe and robust automatic take-off maneuver for a fixed-wing UAV

Abstract: The SAGITTA Demonstrator is a novel UAV with a digital flight control system being developed by the Institute of Flight System Dynamics of the Technical University of Munich. This paper presents the implementation of an automatic take-off algorithm as a system automation module of the auto flight system. Based on a phase-breakdown of take-off, the implementation features a state machine that covers the procedure of the maneuver and the transition conditions from one phase to another. Through this state machine, the automatic take-off algorithm enables controller modules of the auto flight system and provides corresponding commands for the conduct of take-off. The design of the automatic take-off algorithm focuses on maximizing safety and robustness against uncertainties and disturbances. Presented simulation results prove the concept of the algorithm as well as its robustness to simulated reality effects.

Recognition of human activity using Internet of Things in a non-controlled environment

Abstract: Recognition of human activity in a non-controlled environment remained an unsolved problem. In this study, we wonder if accurate recognition of activity can be obtained using Internet of Things technology. We propose a processing methodology that can allow to an automatic system like a robot to recognize human activity. The approach consists of the classification of some activities of the subjects: walking, standing, sitting and laying. The study exploits a standard smartwatch and a smartphone carried by participants during a non-controlled experiment. We propose a pre-computation using Discrete Cosine Transform (DCT), and we identify the best window width and feature length that provides the best results. We show that Support Vector Machines (SVM) provides better results compared with Decision Tree algorithms (DT). The results also demonstrate that participants' activities were classified with an accuracy of more than 91% in a non-controlled environment, with a non-controlled position of the smartphone. We define the notion of transient that corresponds to the transition between two activities as well between two positions of the sensors. The last result shows that removing the transients provides better results for the classification, i.e. 98%. This approach shows that it is possible for a robot assistant to understand the human behavior only using standard Internet of Things technology in a non-controlled environment.

Nonlinear output feedback tracking control of a quadrotor UAV in the presence of uncertainty

Abstract: A nonlinear output feedback control method is presented, which achieves asymptotic altitude and attitude trajectory tracking of a quadrotor system in the presence of model uncertainty and unknown external disturbances. To address a practical scenario where velocity measurements are not available, a bank of dynamic filters is utilized, which acts as a velocity observer in the closed-loop system. A primary focus of the proposed control method is detailed development that shows how a computationally minimal strategy can be utilized to compensate for multiple sources of uncertainty without the use of adaptive parameter estimation or function approximators. A rigorous Lyapunov-based stability analysis is utilized to prove semi-global asymptotic trajectory tracking in the presence of parametric uncertainty and unmodelled external disturbances. Numerical simulation results are also provided to demonstrate the performance of the proposed control law.

Investigation on driver stress utilizing ECG signals with on-board navigation systems in use

Abstract: People today rely more and more on global positioning system (GPS) for navigation when driving in unfamiliar environments. While GPS navigation is indispensable in an intelligent vehicle and provides convenience for road direction, concerns are also raised if the use of GPS may distract drivers to increase unnecessary stress. In this paper, we explore the effects of using GPS navigation on driver stress utilizing electrocardiogram (ECG) signals. In particular, the effects of higher or lower density of GPS instructions are studied. To analyze the driver stress, eight heart rate variability (HRV) features, which were commonly utilized in human stress related studies, were computed from ECG signals. Statistical significance tests were then performed to each HRV feature, so that those effective features for detecting driver stress may be localized. Our studies, based on road driving experiments with ten healthy subjects, showed that MeanRR, SDNN and HRVTri are the top three effective features to detect driver stress, while frequency domain features in general are not sensitive to driver stress. Based on the effective features, our analysis further showed that basically, driving with higher density of GPS instructions has no significant driver stress difference from driving with lower density of GPS instructions.

Model predictive motion control of autonomous forklift vehicles with dynamics balance constraint

Abstract: In this paper, a model predictive control is developed for motion planning and control of nonholonomic autonomous forklifts. The proposed model predictive motion control (MPMC) determines the control inputs for tracking a desired path through minimising path-following error, subject to nonholonomic and dynamic balance constraints of the forklift. MPMC also automatically adjusts the tracking velocity and acceleration to traverse the path as fast as possible without violating constraints while minimising the tracking error. In order to facilitate the real-time implementation of MPMC for industrial applications, a successive online linearisation of the nonlinear discrete-time kinematic model of the forklift is employed along with linear time-varying approximation of the tracking errors. The effectiveness of the proposed method has been demonstrated through simulations.

Monocular urban localization using street view

Abstract: This paper presents a metric global localization in the urban environment only with a monocular camera and the Google Street View database. We fully leverage the abundant sources from the Street View and benefits from its topo-metric structure to build a coarse-to-fine positioning, namely a topological place recognition process and then a metric pose estimation by local bundle adjustment. Our method is tested on a 3 km urban environment and demonstrates both sub-meter accuracy and robustness to viewpoint changes, illumination and occlusion. To our knowledge, this is the first work that studies the global urban localization simply with a single camera and Street View.

Development of a lightweight power-assist suit using pneumatic artificial muscles and balloon-amplification mechanism

Abstract: Low back pain of workers is increased in workplaces that involve hard work. It is caused by excessive loading of the waist joint and muscle fatigue. Therefore, power-assist suits have been developed to decrease the load on the waist joint. However, previous power-assist suits have had problems such as low output or heaviness. Thus, we have developed an assist suit that is lightweight and exerts a large force. A balloon actuator and pneumatic artificial muscles are attached to assist suit as the actuators. The assist suit has various desirable features: lightweight, flexibility, and high output. First, human motion is analyzed for the development of the assist suit. The assist suit is developed to assist the waist joint torque. The assist suit is modeled, and the theoretical values of its generation force are estimated using an assist-suit model. Finally, the assist suit is evaluated by measuring the surface electromyography (EMG). The EMG of the wearer is compared with that without the suit. The effectiveness of the assist suit is confirmed by a decrease in EMG.

A performance evaluation of detectors and descriptors for UAV visual tracking

Abstract: This paper is made up of a series of performance evaluations of computer vision algorithms, namely detectors and descriptors. The OpenCV 3.1 implementations of these algorithms were used for these evaluations. The main purpose behind these evaluations was to determine the best algorithms to use for a UAV guidance system.

Jaya algorithm for solving urban traffic signal control problem

Abstract: This paper studies a large-scale urban traffic signal control problem (LUTSCP). A centralized model is developed for describing the LUTSCP in a scheduling framework. The objective is to minimize the total network-wise delay in a fixed time window. We have implemented a recently developed algorithm, so called Jaya, to solve the LUTSCP. The population initialization is based on the four stages of traffic signal in Singapore. A simple new solution generation strategy is proposed to improve the performance of the Jaya. A neighborhood search operator is proposed based on the characteristics of LUTSCP to improve the search performance in local search space. Experiments are carried out using the traffic signal data from Singapore traffic network. The performance of the new strategy for generating feasible solution and the neighborhood search operator are evaluated and discussed. The optimization results obtained by standard Jaya algorithm and its variants are compared to those by existing traffic signal control system. The comparisons and discussions verify that the Jaya algorithm and its variants are superior over the existing traffic light control. In future work, we will compare the performance of Jaya algorithm to existing intelligent algorithms in literature.

Dynamic flight path control coupling for energy and maneuvering integrity

Abstract: Traditional autopilots suffer from inherent flight path control objective conflict problems, as arbitrary flight path and speed targets cannot be maintained with saturated energy rate control. Elementary flight envelope protections are typically introduced to protect airspeed and prevent loss of control, but rather as a "last line of defense", than as a mean of smooth and deterministic control objective resolution during normal operation. In this paper, an approach for active energy distribution prioritization and integrity protection, as integrated part of the flight path controller of a modular flight guidance and control system, is presented. The approach allows speed or flight path angle maneuvering to be prioritized in case of saturated energy control, with automatic speed priority at the edges of the envelope in order to ensure the airspeed integrity of the aircraft. The approach is analyzed and validated using high-fidelity simulations of the full closed loop system for various conditions.