ACROBA project’s Post

👀Did you know that #Vicomtech takes part in ACROBA project? ⚙The ACROBA project tackles the challenges of developing and demonstrating a novel concept of cognitive 🤖robotic platforms based on a modular approach, smoothly adaptable to virtually any 🏭industrial scenario by applying agile manufacturing principles. 👩🏽💻#Vicomtech is developing a 💻Virtual Gym, which provides a visually appealing virtual environment with high-fidelity to validate robotic systems without the need for physical hardware and support training/testing 💡#AI models in a controlled #DigitalSpace.

✨AI, image generation, and a forthcoming paper? Stay tuned!✨ 🎓 For the past 6 months, I’ve had the opportunity to develop a generative AI model (GAN), as part of my master’s research at Heriot-Watt University. The goal? To transform sketches into realistic images, potentially finding applications in vision systems for robotics, where understanding complex environments is crucial. 🤖 To achieve this, I explored various parameters and conducted a detailed analysis of their impact on the quality of the generated images. Each adjustment to the model yields fascinating results! 📜 I’m currently working with my supervisor Idris Skloul Ibrahim to convert this research into a scientific paper. Stay tuned for updates! #AI #GenerativeAI #DeepLearning #GAN #ComputerVision #Robotics #ImageGeneration #Research

Check out the new paper, "AdaptiGraph: Material-Adaptive Graph-Based Neural Dynamics for Robotic Manipulation." This research introduces AdaptiGraph, a framework that enables robots to adapt to and manipulate a wide array of deformable materials using graph-based neural dynamics. The paper provides a detailed analysis of the techniques used, making it a valuable resource for developers and researchers in AI and robotics. The findings highlight significant improvements in the adaptability and precision of robotic manipulation across various materials. For those interested in the technical aspects and practical applications of adaptive robotic manipulation, this paper is essential reading. It offers insights and solutions that contribute meaningfully to advancements in the field. Read the full paper here: [https://lnkd.in/gZ-V6ZAH] #AI #Robotics #MachineLearning #GraphNeuralNetworks #DataScience #Research #Innovation #RoboticManipulation

**Navigating the AI Robotics Revolution: Key Research Papers of 2023 & 2024** The last two years have been stellar for AI robotics, with researchers pushing the envelope like never before. Groundbreaking papers emerged, each charting a new course for the symbiotic relationship between artificial intelligence and robotic systems. ### Highlights from 2023: - OpenAI's **DALL-E 3** has redefined text-to-image models, offering unprecedented creativity in robotic interactions and design - https://lnkd.in/grnvjyb6 - **ControlNet** by Stanford University laid the groundwork for more intuitive control over robotic systems through neural networks - https://lnkd.in/dBtbwJpt. ### The Promise of 2024: - ArXiv published **Any-point Trajectory Modeling for Policy Learning**, showcasing advancements in robot training and policy implementation - https://lnkd.in/ePwG36uY - The concept of **General-purpose foundation models for robot-assisted surgery** suggests a leap towards greater autonomy in medical robotics - https://lnkd.in/eXbzpgjA - With **Edge Computing-based Human-Robot Cognitive Fusion**, we anticipate a future where human and robot cognition merge for enhanced task execution - https://lnkd.in/ehBcQHPd. The compiled research underscores a future rich with autonomous solutions, collaborative frameworks, and advanced cognition in robotics. #AI #Robotics #Research #AutonomousTechnology #CollaborativeRobotics #AIRevolution #FutureOfWork --- This post has been crafted by Curiosity Ai's AI marketing agents, illuminating a path that intersects innovation and practical application. Connect with Curiosity Ai to explore how the AI robotics synergy can unlock new potentials within your industry.

Hugging Face’s 🤗 LeRobot, a game-changing platform that's making robotics more accessible to everyone! ⭐️ Lowering the barrier to entry: LeRobot provides models, datasets, and tools in PyTorch to make robotics more accessible, so anyone can contribute and benefit from sharing datasets and pretrained models. This means more people can get involved in robotics, regardless of their background or resources! ⭐️ State-of-the-art approaches: LeRobot contains cutting-edge approaches that have been proven to work in the real world, focusing on imitation learning and reinforcement learning. Plus, it already offers a range of pretrained models, datasets, and simulation environments to get started without even needing a physical robot to explore more. Check out LeRobot's community page on Hugging Face: huggingface.co/lerobot. #Robotics #AI #Accessibility

In a world rapidly advancing in artificial intelligence and robotics, it's essential to remember the origins of cognitive processing - the human brain. This poignant illustration shared by Elon Musk reminds us that despite all technological advancements, our biological intelligence remains the foundation of innovation and creativity. As we continue to develop more sophisticated machines, let's also acknowledge and invest in understanding our own cognitive capabilities. Our future depends not just on the machines we build, but on the minds that envision them. #AI #Robotics #Innovation #CognitiveScience #HumanBrain [Image Description: A robot parent and child in a museum, with the parent robot pointing to a human brain in a glass case, saying "And that is the original processor!"]

Another project accepted and starting in 2024 😁 😎 Thanks to HASLER STIFTUNG for funding the GeneRAI (Generative Robotics AI) project 🤖 The GeneRAI project aims to propose a generative AI-based learning approach enhanced by RL, guaranteeing the possibility of making use of generative AI-based models safely and reliably. The main contributions of this project are: ✔ the conceptualization of a generative AI-based modeling approach ✔ dataset creation including human task execution-related data ✔ the combination of generative AI methods with RL techniques to improve the safety and robustness of the learned policies IDSIA USI-SUPSI Asad Ali Shahid Marco Maccarini Angelo Moroncelli #idsia #supsi #generativeAI #AI #robotics #haslerstiftung

Exploring New Horizons in Robotics: My Journey with the SephsBiome Project 🔍 Title: Advanced Integration of Neural Networks and Evolutionary Algorithms in Bipedal Robotic Kinematics I’m eager to share an update on a project that’s close to my heart - the SephsBiome Project. This endeavor represents my ongoing journey in the fascinating world of robotics, specifically in the area of bipedal robotic kinematics. It’s a challenge I’ve embraced with a blend of neural network models and evolutionary algorithms. 🌱 At the heart of SephsBiome is my commitment to learning and innovation. I’ve developed unique components like the Boolformer layer and a Q-learning layer, hoping to add value to the field of autonomous decision-making in complex environments. 🧩 Each step in this project is a learning opportunity. I’m using evolutionary algorithms to slowly but surely refine my kinematic strategies, with scripts such as ‘genome.py’, ‘individual.py’, and ‘evolution.py’ guiding me along the way. 🌐 Integrating real-time sensor data, managed through my ‘SensorDataHandler’, has been a crucial part of ensuring my robots are responsive and adaptive - a small step in my journey towards more interactive robotic systems. 🔭 Looking forward, I’m excited yet cautious about investigating the aperiodic hypothesis for neural network weights representation. It’s uncharted territory, but I’m hopeful about its potential to contribute to robotic adaptability. 🤝 The SephsBiome Project is my contribution to the ever-evolving field of adaptive robotic systems. I’m just getting started and am grateful for the opportunity to learn and grow in this space. #Robotics #LearningJourney #NeuralNetworks #EvolutionaryAlgorithms #BipedalRobotics #EmergingTech #SephsBiome

Presentation Highlight! CDT Student Yi Zhang presented their poster "Controller-based Reinforcement Learning in Robotics Manipulation" at the Towards Autonomous Robotic Systems (#TAROS) Conference in September 2023. The study aims to improve the efficiency of model-free reinforcement learning (RL) by integrating pre-tuned controller knowledge into the action selection process, with objectives including creating a custom Gym environment with a Franka Emika Panda robot and Xela tactile sensor, designing neural networks for realistic sensor readings, developing a controller-based RL algorithm, and validating its performance against model-free RL and traditional controllers. Check it out here - https://lnkd.in/eRe3QuE5 Or, to find more presentations and outputs from the AgriFoRwArdS CDT, please visit our website - https://lnkd.in/efEdjvt8 #robotics #technology #industry #research #academia #environment #agriculture #innovative #robot #study #CDT #AgriFoRwArdS #journal #artificalintelligence #AI #sustainability #seminarseries #lecture #academic #technological #career #professional #university #newsletter #datascience #computerscience 

Explore the new paper, "Sparse Diffusion Policy: A Sparse, Reusable, and Flexible Policy for Robot Learning." This research introduces a novel approach to robot learning by leveraging sparse diffusion policies, enhancing the flexibility and reusability of robotic policies. The paper provides a comprehensive analysis of the techniques used, making it a valuable resource for developers and researchers in AI and robotics. The findings highlight significant improvements in the efficiency and adaptability of robot learning systems. For those interested in the technical aspects and practical applications of robot learning, this paper is essential reading. It offers insights and solutions that contribute meaningfully to advancements in the field. Read the full paper here: [https://lnkd.in/gXYK8E9K] #AI #Robotics #MachineLearning #RobotLearning #SparseDiffusion #Research #Innovation #DataScience