Robotics Australia Group’s Post

We still have quite some positions available in the Humanoid Sensing and Perception group at the Italian Institute of Technology (https://hsp.iit.it), topics are: • Robot perception and manipulation for social rehabilitation purposes (https://lnkd.in/dfbAc2eA) • Vision based control for intelligent prosthetic hands (https://lnkd.in/dHcZ9wh3) • Software Engineer for Robotics (https://lnkd.in/dHtefNPg) • Control of manipulation for collaborative tasks (https://lnkd.in/drF3YA4j) Please apply online following the links above. #humanoidsrobotics #artificialintelligence #research

Three Days of Cutting-Edge Experimentation with Acoustic Sensing for Underwater Navigation 🌊 The past 3 days have been a whirlwind of discovery, hands-on learning, and technical challenges as I worked with a comprehensive acoustic sensor system designed for the autonomous navigation of underwater vehicles. Here’s what this intensive experimentation involved: 🔍 Sensor Functionality Testing: I spent hours developing and testing programs to record and analyze sensor data, extracting critical parameters like: • Position • Range • Bearing • Localisation 📊 Data Visualization with ROS & Python Coding: To understand the sensor’s performance in real-time, I utilized Robot Operating System (ROS) and Python for data monitoring and visualization. This helped bridge the gap between raw sensor data and actionable insights. 🌏 Real-World Application at Maithon Power Limited (Jv Of Tata Power & Dvc) Dam: Taking the system into a real terrain like Maithon Dam allowed me to see acoustic principles in action and understand how environmental factors like electrical noise, mechanical dynamics, and underwater kinematics interplay. 💡 Multidisciplinary Learning: This wasn’t just about acoustics—it was about how diverse domains seamlessly integrate to make systems like this work. This opportunity was made possible by the CSIR-Central Mechanical Engineering Research Institute (CMERI), Underwater Robotics and Autonomous Systems Group (URASG), under the incredible guidance and mentorship of: • Dr. Sambhunath Nandy (Senior Principal Scientist and Head of URASG Department) • Dr. Siva Ram Krishna Vadali (Senior Principal Scientist) • Dr. Srinivasan Aruchamy (Principal Scientist) • Dr. veeraragavan s (Scientist) • Mr. Akash Pandey (Project Associate) A special thanks to my colleague Surya Prasath S K, whose support throughout this experiment was invaluable. Collaborating with skilled peers always makes the journey more rewarding! Working alongside such experts and mentors from diverse fields has been an eye-opening experience, offering me a level of professional and technical growth that few other opportunities could provide. Here’s to leveraging cutting-edge tech for deeper underwater exploration and pushing the boundaries of autonomous systems! 🌐 #UnderwaterNavigation #AcousticSensing #AutonomousSystems #UnderwaterRobotics #DataVisualization #ROSMelodic #PythonProgramming #MaithonDam #CSIRCMERI #RoboticsResearch #SensorTechnology #MultidisciplinaryEngineering #AcousticsInAction #RealTimeData #EngineeringInnovation #AutonomousVehicles #TechForExploration

🎓✨ Today marks a significant milestone in my journey as I’ve successfully defended my master’s thesis on advanced deep learning semantic segmentation for maritime corrosion detection! 🌊🤖Diving deep into the intersection of electrical engineering and robotics, I explored innovative ways to leverage AI in safeguarding our maritime infrastructure. This project wasn't just about algorithms; it represented a fusion of technology and environmental stewardship, aiming to preserve the oceans we depend on.I’m grateful for the insights from my professors, the encouragement of my peers, and the support of my family!#DeepLearning #MaritimeTech #Robotics #CorrosionDetection #Innovation #SustainableEngineering

🚀 Introducing Expanding Path RRT* (EP-RRT*) - A Leap in Path Planning! 🌟 Path planning lies at the heart of robotics and autonomous systems, enabling efficient navigation through complex environments. We're thrilled to share our latest work on Expanding Path RRT*, an innovative extension of the traditional RRT* algorithm, designed to tackle challenging scenarios with enhanced performance. 🔑 Key Features of EP-RRT*: Heuristic Sampling: Inspired by RRT*-Connect, this approach focuses exploration where optimal paths are more likely. Path Expansion: Strategically grows promising path segments, excelling in narrow or cluttered spaces. Improved Convergence: Faster and more efficient in achieving high-quality paths. Asymptotic Optimality: Retains the rigor of RRT* while enhancing real-world applicability. 📊 Experimental Success: Our algorithm was rigorously tested across various environments: 1️⃣ Open Spaces: Demonstrated general efficiency. 2️⃣ Dense Obstacles: Excelled in urban-like scenarios. 3️⃣ Narrow Corridors: Proved its worth in constrained spaces. 4️⃣ Complex Mazes: Surpassed traditional methods in challenging terrains. 💡 Impact: EP-RRT* is a game-changer for autonomous navigation, search and rescue missions, and industrial robotics, delivering faster solutions and superior paths in intricate environments. 📷 Visual Results & Report: All the experimental image results and the full project report can be found in the GitHub repository 📚 Check out our GitHub repository for more details: https://lnkd.in/eaw8isz6 🌟 Acknowledgments: A big thank you to my incredible teammate, Dhairya Shah, for their dedication and collaboration throughout this project! ✨ Open to Opportunities: I am actively seeking roles in robotics, autonomous systems, and path planning. If you’re hiring or know of any opportunities in this field, let’s connect! #Robotics #PathPlanning #AutonomousSystems #Innovation #EP_RRT*#RRT* #HeuristicAlgorithms #JobSearch #CareerInRobotics #AutonomousNavigation #MachineLearning #AIInRobotics #ResearchAndDevelopment #EngineeringJobs #Collaboration

It was an honor to host the Fall 2024 Workshop on Autonomy in the Gen AI era at the USC Center for Autonomy and AI. We all got to learn from the many thought leaders in academia and industry in this space. It was quite apparent that while organizations in various industrial sectors are grappling in their own ways to develop autonomy technology, the challenges at some fundamental level faced by each of them are all common. There are still challenges in ensuring that AI-based methods can ensure physical safety, operational assuredness, resilience and robustness. At the same time autonomous systems technology, long a goal for decades is a lot closer to realization than ever before. A key challenge is collating training data: It is simply prohibitive to collect large scale datasets needed to train AI models for autonomy. So, Gen AI methods have a key role to play. At the same time, there is potential to exploit cross-domain datasets coupled with autonomy-specific architectural innovations to build AI-Autonomy models that are highly capable across a variety of use cases. A key issue is how to adapt the base AI models for specific use cases. RLHF techniques used in large language and multi-modal models need further development so that they are sample-efficient and can make post-training tenable. Thank you to all the speakers, and all the attendees. We hope to continue some of these conversations, and hopefully some of them will lead to even greater engagement soon.  

Researchers at Université Libre de Bruxelles have developed a swarm architecture inspired by the human nervous system to enhance robot team cooperation. This framework allows robots to self-organize into sub-swarms, improving coordination in tasks like disaster response and pollution monitoring. The architecture combines centralized and self-organized control, forming a dynamic ad-hoc control network, or self-organizing nervous system (SoNS). Robots communicate with direct neighbors, preventing communication bottlenecks. This system acts as middleware, enabling robots to function collectively while maintaining scalability and flexibility. Tested in simulations and real-world experiments, the framework shows promise for real-world applications. Future research aims to enhance SoNS brains and hierarchical computation for online learning and autonomous mission planning.

Check out our conf. paper titled: Deep Reinforcement Learning for Autonomous Satellite Responsiveness to Observed Events. We've developed a DRL agent that is able to react in real time to observations made on orbit in EO missions. Thanks to the collective effort of the team @Kedarnaik, Clayton Kotulak, and Andrew Harris at BAE Systems, Inc. #DRL #AIinSpace #Autonomy

Join us tomorrow, 10th April at 2pm at for Louise Dennis inaugural lecture on Verifiable Autonomous Systems at The University of Manchester, Kilburn building Lecture Theatre. 1.3   How can we provide guarantees of behaviour for #autonomous systems? In this talk I will explain how autonomous systems can be programmed in ways that make them amenable to formal #verification. I will discuss how cognitive agent programming can be used to provide a transparent and verifiable decision-making core and show how properties of the system can then be checked – including properties related to #ethical behaviour.   cc CRADLE Robotics and AI https://lnkd.in/erhtS8xs

🚀 Exciting strides in autonomous laboratory technology! Check out our latest research published in Nature Communications: [https://lnkd.in/gf_iFgFe] Traditional autonomous labs are limited by single-user environments, which restricts usability of lab resources. To address this, I developed a OS for autonomous labs, OCTOPUS, enabling multi-user environments and facilitating remote experimentation. In multi-user environments, parallelizing numerous experiments can be challenging. That’s why we designed a user-optimal scheduler within OCTOPUS, reducing experiment times significantly—from 13 hours to just 2 hours!⏱️ To support researchers starting with autonomous labs, we also introduced Copilot of OCTOPUS with GPT. This system simplifies code generation and adapts regardless of any other applications, offering high scalability for researchers in autonomous lab field. Excited to see how these advancements can propel multi-user autonomous labs forward in diverse scientific domains!