Coordinating a team of mobile robots that simultaneously act in a shared environment is a challenging problem that comes in different flavors. One of the most widely studied abstractions of it is known as Multi-Agent Path Finding (MAPF), which adopts several simplifying assumptions regarding how the agents move, communicate, and observe the environment. Heuristic search is a widely used technique to obtain high-quality MAPF solutions. However, search-based solvers often need to scale better with the number of agents. To mitigate this issue and to lift several other restrictions, decentralized, learning-based approaches to MAPF recently came on stage. They utilize the power of modern machine learning, to obtain decision-making policies that do not require a centralized controller, work well under partial observability and limited communication, and thus are potentially more suited to large-scale robotic applications despite the lack of theoretical guarantees. In this tutorial, we propose to overview the core problem of multi-robot pathfinding and summarize recent progress in the field. Our objective is to give a holistic perspective covering theoretical background, practical algorithms, and software needed to create modern learnable MAPF solvers. As the tutorial involves demo sessions with (some) coding in Python, basic knowledge of this language is beneficial.
In today's rapidly evolving landscape of cyber-security threats and the widespread adoption of nano-scale devices, intelligent camera-based functionalities within smart cyber-physical systems (CPS) and the Internet-of-Things (IoT) encounter unprecedented challenges. These challenges stem from emerging attack vectors and security/privacy risks associated with the processing of image and video data. Beyond traditional concerns such as IP theft and data breaches, modern machine learning (ML) systems operating on visual data face significant adversarial and backdoor threats. Adversarial and backdoor attacks involve deliberate manipulations in images, exploiting vulnerabilities inherent in machine/deep learning models and learning mechanisms. These attacks can severely compromise system performance and decision-making processes. Addressing these evolving security and privacy threats necessitates continual advancements in defense and obfuscation strategies. These strategies play a crucial role in fortifying the resilience of intelligent systems deployed across diverse image and video processing applications, including computer vision. Through hands-on demonstrations and practical examples, attendees of the tutorial will gain insights into effectively defending against adversarial and backdoor attacks. These attacks target inherent vulnerabilities in ML models and learning mechanisms used for tasks such as depth estimation, object detection, and classification. Additionally, the tutorial will explore emerging threats specific to autonomous systems and mobile robots, offering strategies to safeguard these systems against evolving security and privacy risks.
Drones have witnessed a remarkable increase in their usage across various disciplines, including agriculture, surveillance, healthcare, forestry, and military domains. The current international market of surveillance and rescue drones is 3254.4 Million US$ and it will potentially grow to 11,468 Million US$ till 2033. Most drones are outfitted with sensors and cameras, enabling the creation of specific applications/solutions across various domains. With the rapid advancement in drone technology and AI, there is an urgent need for professionals to stay abreast of the latest developments in real-time AI applications for drones. As drones continue to evolve and play a pivotal role in various industries, AI-powered visual data processing becomes imperative for extracting meaningful information/insights from the vast amount of data collected by drones including AI-on-the-Edge perspective. Keeping in view the rapidly evolving landscape of drone technology and the pressing need for advanced real-time AI-based data processing, we have designed this tutorial to acquaint the audience with the latest tools, methods, and research trends in Edge AI applications for drone technology. As the tutorial delves into advanced methodologies, practical implementations, and real-world applications, participants will concurrently explore the ethical considerations and responsible AI practices crucial for building trustworthy drone AI systems.
Join the tutorial section of “From Hover to Horizon: Mastering Drone Control in MATLAB”. This tutorial will deep dive into the design and tuning of control systems for drones, with a focus on both multirotor and fixed-wing types, using MATLAB® and Simulink®. By broadening the scope from VTOL UAVs to encompass drones in general, participants will explore a wide array of control challenges and solutions applicable to various drone configurations. The session will guide attendees through a structured approach using a MATLAB Project reference application template, covering the fundamentals of drone behavior, control system design, tuning strategies for different flight modes, and deployment to hardware for real-world application. This tutorial targets a broad audience, including engineers, researchers, and students. While a foundational understanding of control theory and familiarity with MATLAB and Simulink are advantageous, the session is designed to accommodate a range of expertise levels.
This tutorial aims to explore the immense potential of simulation in the field of robotics. Simulation has emerged as a powerful tool for testing and validating various robotics algorithms and systems, significantly reducing costs and time associated with real-world experimentation. This tutorial will delve into the fundamentals of simulation in robotics, covering topics such as simulation platforms, modeling techniques, sensor simulation, and motion planning. This will include control techniques for autonomous vehicles such as aerial robots and wheeled robots. Practical demonstrations and case studies will illustrate the effectiveness of simulation in accelerating robotics research and development. This tutorial will also demonstrate the real-world industry level usage of the simulations.
In recent decades, the rise of contact robots has transformed fields from personalized assistance to rehabilitation. These robots handle intricate tasks like team collaboration and task-oriented motion, demanding control mechanisms ensuring safe, intuitive, and efficient human interactions. This tutorial delves into advanced control strategies applicable to both robotic systems and modeling human sensorimotor functions. Key topics include nonlinear stochastic optimal control, which models human motor behaviors accounting for variability and sensory noise. Haptic communication strategies derived from cooperative tasks enhance performance and learning through prediction and adaptation. Passivity theory regulates energy exchanges in interactive systems, while differential game theory crafts cooperative strategies for participants with diverse roles. This comprehensive framework, including our significant contributions, facilitates understanding and modeling of sensorimotor interactions, designing optimal human-robot interaction strategies, and coordinating multiple robots. Participants will gain insights into applying and extending these techniques across diverse robotics and human-centric applications, shaping the future of interactive robotics.
This tutorial will provide a comprehensive introduction to Intrepid AI, an innovative platform designed to streamline the development of autonomous robotics applications. Participants will first explore the platform's core features, including its integrated editor, simulator, and deployment mechanism, which collectively empower robotics designers to create sophisticated behaviors efficiently. Attendees will then delve into the platform's simulation capabilities, learning how to visualize and validate their robotic designs in a virtual environment before real-world deployment. The tutorial will feature a hands-on session introducing the Intrepid graph, a user-friendly, low-code interface that simplifies the creation and modification of complex robotic behaviors, making advanced robotics accessible to both experts and non-experts. The session will culminate with the announcement of the Intrepid Challenge, an engaging competition that allows participants to apply their newfound skills in a collaborative and competitive setting. This tutorial promises to equip attendees with practical knowledge and skills, demonstrating how Intrepid AI can revolutionize the process of developing autonomous robotics solutions from conception to deployment.
Foundations of Interaction Control for Contact Robots: Energy-based Methods and Interactive Learning
In recent decades, the proliferation of contact robots—designed to physically interact with humans—has revolutionized applications ranging from personalized assistance to physical training and rehabilitation. These robots, which engage in complex tasks such as facilitating teamwork or providing resistance in task-oriented movements, require control mechanisms that ensure safe, intuitive, and efficient interactions with both their human users and the environment. Designing such optimal interactions necessitates an understanding of human sensorimotor behavior and the use of advanced control strategies. This tutorial on "Foundations of interaction control for contact robots" will thus include the following key topics: Nonlinear stochastic optimal control for modeling of human motor behaviors, considering the inherent variability and noisy sensory signals. Haptic communication strategies based on observations of human cooperative tasks, emphasizing prediction and adaptation to enhance performance and learning. The application of passivity theory to manage energy exchanges in physically interactive systems.Differential game theory for crafting cooperative strategies among participants with distinct roles.This body of work, to which we have significantly contributed, forms a comprehensive framework for understanding and modeling sensorimotor interactions between humans, designing intuitive and optimal human-robot interaction strategies, and coordinating multiple robots. Participants will gain insights into these techniques and learn how to apply and expand upon them in various robotics and human-centric applications.
Robot planning and Control
14th October (09:00-12:00)
Neurorobotics; Motion Control; Sensorimotor Learning
Meeting Room 1
Etienne burdet, Sami Haddadin, Abdalla Swikir*, Erfan Shahriari
In recent decades, the proliferation of contact robots—designed to physically interact with humans—has revolutionized applications ranging from personalized assistance to physical training and rehabilitation. These robots, which engage in complex tasks such as facilitating teamwork or providing resistance in task-oriented movements, require control mechanisms that ensure safe, intuitive, and efficient interactions with both their human users and the environment. Designing such optimal interactions necessitates an understanding of human sensorimotor behavior and the use of advanced control strategies. This tutorial on "Foundations of interaction control for contact robots" will thus include the following key topics: Nonlinear stochastic optimal control for modeling of human motor behaviors, considering the inherent variability and noisy sensory signals. Haptic communication strategies based on observations of human cooperative tasks, emphasizing prediction and adaptation to enhance performance and learning. The application of passivity theory to manage energy exchanges in physically interactive systems.Differential game theory for crafting cooperative strategies among participants with distinct roles.This body of work, to which we have significantly contributed, forms a comprehensive framework for understanding and modeling sensorimotor interactions between humans, designing intuitive and optimal human-robot interaction strategies, and coordinating multiple robots. Participants will gain insights into these techniques and learn how to apply and expand upon them in various robotics and human-centric applications.
We propose a competition in humanoid sprint in simulation with the transfer to the real hardware. Robust trajectory following for walking remains a substantial challenge in humanoid robotics. We have created a system for simulation and automatic performance assessment for multiple competitions, including humanoid sprint and marathon. It relies on a physics simulator and rules that closely follow FIRA’s guidelines.
The multi-award winning Abu Dhabi National Exhibition Centre (ADNEC) is the largest exhibition venue in the Middle East, fully interconnected within the hearth of Abu Dhabi. Featuring a unique architectural design, ADNEC offers one of the world's most accessible and flexible event spaces.