Invited Speakers:
Kostas Alexis
Professor
Professor
Department of Engineering Cybernetics
Norwegian University of Science and Technology
Fei Gao
Tenured Associate Professor
Tenured Associate Professor
Department of Control Science and Engineering
Zhejiang University
Yang Lyu
Associate Professor
Associate Professor
School of Automation
Northwestern Polytechnical University
Hannibal Paul
Assistant Professor
Assistant Professor
Department of Robotics
College of Science and Engineering
Ritsumeikan University
Guillaume Sartoretti
Assistant Professor
Assistant Professor
Department of Mechanical Engineering
National University of Singapore
Boyu Zhou
Assistant Professor
Assistant Professor
School of Artificial Intelligence
Sun Yat-sen University
Workshop Program (GMT+4)
Opening and welcome
Multi-Source Information Fusion for Reliable Unmanned System Localization in GNSS Denied Environments
Abstract: Achieving reliable and persistent localization in GNSS denied environment is vital for many safety-critical unmanned systems. In this report, we report our recent works fuse multi-source information to achieve higher quality of navigation results. Especially, we explore 3 levels of fusion strategy, namely the sensor-level, feature level and coordinate level, for different unmanned system applications, such as quadrotors and eVTOL UAVs
Autonomous Multi-modal Ship Ballast Tank Inspection
Abstract:
This talk focuses on technologies that enable the autonomous
inspection of ship ballast tanks without assuming access to any
information a priori. Critically, the presented results exploit a real-time
inferred semantic understanding of the environment which allows the onboard autonomy
to reason about the structures within ballast tanks and predict their geometric configuration
thus greatly reducing the time necessary for inspection. Extensive results will be presented
from real-life deployments of aerial robots in ballast water tanks.
UAV Systems for Aerial Manipulation: Consulting on Inspection Applications in Hard-to-Reach Environments
Abstract:This talk will focus on UAV systems developed for aerial manipulation tasks, with an emphasis on inspection applications in hard-to-reach environments. The presented UAVs are capable of performing inspection operations in various scenarios by leveraging advanced control strategies and adaptable configurations. Approaches for aerial manipulation will be discussed, highlighting how these systems can be utilized in different scenarios where conventional methods are often limited. Key experimental results and insights will be shared to illustrate the potential of these UAVs in enhancing both manipulation and inspection processes.
Coffee break and Poster Session
Towards Learned Cooperation at Scale in Robotic Multi-Agent Systems
Abstract: With the recent advances in sensing, actuation, computation, and communication, the deployment of large numbers of robots is becoming a promising avenue to enable or speed up complex tasks in areas such as manufacturing, last-mile delivery, search-and-rescue, or autonomous inspection. My group strives to push the boundaries of multi-agent scalability by understanding and eliciting emergent coordination/cooperation in multi-robot systems as well as in articulated robots (where agents are individual joints). Our work mainly relies on distributed (multi-agent) reinforcement learning, where we focus on endowing agents with novel information and mechanisms that can help them align their decentralized policies towards team-level cooperation. In this talk, I will first summarize my early work in independent learning, before discussing my group's recent advances in convention, communication, and context-based learning. I will discuss these techniques within a wide variety of robotic applications, such as multi-agent path finding, autonomous exploration/search, task allocation, and legged locomotion. Finally, I will also touch on our recent incursion into the next frontier for multi-robot systems: cooperation learning for heterogeneous multi-robot teams. Throughout this journey, I will highlight the key challenges surrounding learning representations, policy space exploration, and scalability of the learned policies, and outline some of the open avenues for research in this exciting area of robotics.
Active Perception and Mobile Manipulation with Autonomous UAVs
Abstract: In recent years, unmanned aerial vehicles (UAVs)
have garnered significant attention due to their high flexibility and mobility.
This talk will present research on UAVs in the areas of active perception and mobile manipulation,
with applications in inspection and logistics. First, we will introduce methods for autonomous drones
to efficiently explore unknown environments, including advancements in real-time planning,
efficient environmental representation, and swarm collaboration. Next, we will discuss
the challenges of coverage and reconstruction in complex 3D scenes,
presenting prediction-enhanced real-time coverage planning methods and
heterogeneous drone collaboration strategies. Finally, we will explore
recent advancements in UAV-based transportation, delivery, and manipulation.
State Estimation and Planning for Aerial Swarm
Abstract: In recent years, our community has witnessed huge development in autonomous navigation. From the reliance on GPS and external planning, to the fully onboard perception and computation, aerial robots and their swarms are now flying out of the laboratory and are capable of performing simple tasks. However, constrained by their perception range, computational capacity, and maneuverability, swarm aerial robotic systems face challenges such as degradation in harsh environments, divergence in swarm positioning, and failure in large-scale maneuver planning. In this talk, I will introduce some new methods developed by my group, in areas such as novel dynamic vision with active texture enhancement, robust state estimation under degradation, dynamic environmental perception, ultra-lightweight obstacle-free space abstraction, and high-fidelity 3D reconstruction. Then, based on real-world requirements, some systematic solutions for specific tasks are presented, where the architecture, algorithms, engineering considerations, as well as closed-loop performances, are explained. Finally, I will turn to some of our most recent research, on which we are working towards a perception-planning coupled, flexibly coordinated, and large-scale aerial swarm system.
Champion Solution
Abstract: Unmanned aerial vehicles (UAVs) can be employed for specialized operations such as environmental exploration and inspection tasks. Compared to a single drone or a homogeneous fleet, a fleet of heterogeneous, cooperative UAVs can better balance time and financial costs. To address this challenge, we propose an efficient solution to CARIC at IROS 2024. In our approach, explorer drones equipped with a high-performance LiDAR will focus on exploring the environment. Implementing a moving-to-obstacle policy, explorers sample waypoints both within and around a bounding box that contains the interest points, visit them sequentially following a heuristic rule that indicates the priority of the waypoint, and provide maps while identifying interest points for lightweight photographer drones. Utilizing this information, the photographers can cluster the known interest points and divide the inspection tasks. By visiting each category of interest points in batches, the photographers can perform more detailed inspections of these points and their surroundings to improve the inspection results. The drones navigate using the A* algorithm with a distance-based heuristic function, and the trajectories are further optimized with the minimal-snap method. The effectiveness and robustness of our method have been shown in the CARIC.