Marco Zennaro

Bio: Marco Zennaro is an academic researcher from International Centre for Theoretical Physics. The author has contributed to research in topics: Wireless sensor network & White spaces. The author has an hindex of 24, co-authored 161 publications receiving 2252 citations. Previous affiliations of Marco Zennaro include Royal Institute of Technology & University of California, Berkeley.

An overview of emerging results in cooperative UAV control

Abstract: Inexpensive fixed wing UAV are increasingly useful in remote sensing operations. They are a cheaper alternative to manned vehicles, and are ideally suited for dangerous or monotonous missions that would be inadvisable for a human pilot. Groups of UAV are of special interest for their abilities to coordinate simultaneous coverage of large areas, or cooperate to achieve goals such as mapping. Cooperation and coordination in UAV groups also allows increasingly large numbers of aircraft to be operated by a single user. Specific applications under consideration for groups of cooperating UAV are border patrol, search and rescue, surveillance, communications relaying, and mapping of hostile territory. The capabilities of small UAV continue to grow with advances in wireless communications and computing power. Accordingly, research topics in cooperative UAV control include efficient computer vision for real-time navigation and networked computing and communication strategies for distributed control, as well as traditional aircraft-related topics such as collision avoidance and formation flight. Emerging results in cooperative UAV control are presented via discussion of these topics, including particular requirements, challenges, and some promising strategies relating to each area. Case studies from a variety of programs highlight specific solutions and recent results, ranging from pure simulation to control of multiple UAV. This wide range of case studies serves as an overview of current problems of Interest, and does not present every relevant result.

Large-scale privacy protection in Google Street View

Abstract: The last two years have witnessed the introduction and rapid expansion of products based upon large, systematically-gathered, street-level image collections, such as Google Street View, EveryScape, and Mapjack. In the process of gathering images of public spaces, these projects also capture license plates, faces, and other information considered sensitive from a privacy standpoint. In this work, we present a system that addresses the challenge of automatically detecting and blurring faces and license plates for the purpose of privacy protection in Google Street View. Though some in the field would claim face detection is “solved”, we show that state-of-the-art face detectors alone are not sufficient to achieve the recall desired for large-scale privacy protection. In this paper we present a system that combines a standard sliding-window detector tuned for a high recall, low-precision operating point with a fast post-processing stage that is able to remove additional false positives by incorporating domain-specific information not available to the sliding-window detector. Using a completely automatic system, we are able to sufficiently blur more than 89% of faces and 94 – 96% of license plates in evaluation sets sampled from Google Street View imagery.

Strategies of Path-Planning for a UAV to Track a Ground Vehicle

Abstract: In this paper, we present a strategy of path-planning for an unmanned aerial vehicle (UAV) to follow a ground vehicle. The ground vehicle may change its heading and vary its speed from a standstill up to the velocity of the UAV, while the UAV will maintain a fixed airspeed and will maneuver itself to track the ground vehicle. The algorithm also allows the UAV to track the ground vehicle with an offset vector (i.e. the user may wish the UAV to stay ahead of the ground vehicle or to its sides). Since the ground vehicle may operate in a range of velocities, the algorithm must plan the UAV's path with the appropriate schemes for the various ground vehicle speeds. The natural effect of wind injects a disturbance into the system, and so wind compensation techniques had to be developed. In order to maintain the focus of this project on path-planning strategies, the path-planning algorithm was implemented on top of a system that already controls the dynamics of the UAV. Simulation of aircraft and ground vehicles was performed with a hardware-in-the-loop simulation environment to test for mission feasibility. After attaining satisfactory simulation results, an experiment was conducted to confirm the path-planning strategy.

On the design of smart parking networks in the smart cities: an optimal sensor placement model

Abstract: Smart parking is a typical IoT application that can benefit from advances in sensor, actuator and RFID technologies to provide many services to its users and parking owners of a smart city. This paper considers a smart parking infrastructure where sensors are laid down on the parking spots to detect car presence and RFID readers are embedded into parking gates to identify cars and help in the billing of the smart parking. Both types of devices are endowed with wired and wireless communication capabilities for reporting to a gateway where the situation recognition is performed. The sensor devices are tasked to play one of the three roles: (1) slave sensor nodes located on the parking spot to detect car presence/absence; (2) master nodes located at one of the edges of a parking lot to detect presence and collect the sensor readings from the slave nodes; and (3) repeater sensor nodes, also called “anchor” nodes, located strategically at specific locations in the parking lot to increase the coverage and connectivity of the wireless sensor network. While slave and master nodes are placed based on geographic constraints, the optimal placement of the relay/anchor sensor nodes in smart parking is an important parameter upon which the cost and efficiency of the parking system depends. We formulate the optimal placement of sensors in smart parking as an integer linear programming multi-objective problem optimizing the sensor network engineering efficiency in terms of coverage and lifetime maximization, as well as its economic gain in terms of the number of sensors deployed for a specific coverage and lifetime. We propose an exact solution to the node placement problem using single-step and two-step solutions implemented in the Mosel language based on the Xpress-MPsuite of libraries. Experimental results reveal the relative efficiency of the single-step compared to the two-step model on different performance parameters. These results are consolidated by simulation results, which reveal that our solution outperforms a random placement in terms of both energy consumption, delay and throughput achieved by a smart parking network.

LPWAN Technologies: Emerging Application Characteristics, Requirements, and Design Considerations

Abstract: Low power wide area network (LPWAN) is a promising solution for long range and low power Internet of Things (IoT) and machine to machine (M2M) communication applications. This paper focuses on defining a systematic and powerful approach of identifying the key characteristics of such applications, translating them into explicit requirements, and then deriving the associated design considerations. LPWANs are resource-constrained networks and are primarily characterized by long battery life operation, extended coverage, high capacity, and low device and deployment costs. These characteristics translate into a key set of requirements including M2M traffic management, massive capacity, energy efficiency, low power operations, extended coverage, security, and interworking. The set of corresponding design considerations is identified in terms of two categories, desired or expected ones and enhanced ones, which reflect the wide range of characteristics associated with LPWAN-based applications. Prominent design constructs include admission and user traffic management, interference management, energy saving modes of operation, lightweight media access control (MAC) protocols, accurate location identification, security coverage techniques, and flexible software re-configurability. Topological and architectural options for interconnecting LPWAN entities are discussed. The major proprietary and standards-based LPWAN technology solutions available in the marketplace are presented. These include Sigfox, LoRaWAN, Narrowband IoT (NB-IoT), and long term evolution (LTE)-M, among others. The relevance of upcoming cellular 5G technology and its complementary relationship with LPWAN technology are also discussed.

Reading Digits in Natural Images with Unsupervised Feature Learning

Abstract: Detecting and reading text from natural images is a hard computer vision task that is central to a variety of emerging applications. Related problems like document character recognition have been widely studied by computer vision and machine learning researchers and are virtually solved for practical applications like reading handwritten digits. Reliably recognizing characters in more complex scenes like photographs, however, is far more difficult: the best existing methods lag well behind human performance on the same tasks. In this paper we attack the problem of recognizing digits in a real application using unsupervised feature learning methods: reading house numbers from street level photos. To this end, we introduce a new benchmark dataset for research use containing over 600,000 labeled digits cropped from Street View images. We then demonstrate the difficulty of recognizing these digits when the problem is approached with hand-designed features. Finally, we employ variants of two recently proposed unsupervised feature learning methods and find that they are convincingly superior on our benchmarks.

Computer vision : a modern approach = 计算机视觉 : 一种现代的方法

Abstract: From the Publisher: The accessible presentation of this book gives both a general view of the entire computer vision enterprise and also offers sufficient detail to be able to build useful applications. Users learn techniques that have proven to be useful by first-hand experience and a wide range of mathematical methods. A CD-ROM with every copy of the text contains source code for programming practice, color images, and illustrative movies. Comprehensive and up-to-date, this book includes essential topics that either reflect practical significance or are of theoretical importance. Topics are discussed in substantial and increasing depth. Application surveys describe numerous important application areas such as image based rendering and digital libraries. Many important algorithms broken down and illustrated in pseudo code. Appropriate for use by engineers as a comprehensive reference to the computer vision enterprise.

nuScenes: A multimodal dataset for autonomous driving

Abstract: Robust detection and tracking of objects is crucial for the deployment of autonomous vehicle technology. Image based benchmark datasets have driven development in computer vision tasks such as object detection, tracking and segmentation of agents in the environment. Most autonomous vehicles, however, carry a combination of cameras and range sensors such as lidar and radar. As machine learning based methods for detection and tracking become more prevalent, there is a need to train and evaluate such methods on datasets containing range sensor data along with images. In this work we present nuTonomy scenes (nuScenes), the first dataset to carry the full autonomous vehicle sensor suite: 6 cameras, 5 radars and 1 lidar, all with full 360 degree field of view. nuScenes comprises 1000 scenes, each 20s long and fully annotated with 3D bounding boxes for 23 classes and 8 attributes. It has 7x as many annotations and 100x as many images as the pioneering KITTI dataset. We define novel 3D detection and tracking metrics. We also provide careful dataset analysis as well as baselines for lidar and image based detection and tracking. Data, development kit and more information are available online.