Pulkit Agrawal’s Post

Have you ever wondered why there are so few intelligent robots deployed in factories, despite strong progress on robotic AI in academic and corporate labs worldwide? Our newest paper (link below 👇) charts a path toward closing the AI application gap in industrial robotics and proposes concrete methods for deploying intelligent robots to the shopfloor economically and at scale 📈 🚀 In "BANSAI: Towards Bridging the AI Adoption Gap in Industrial Robotics with Neurosymbolic Programming", Julia Dvorak, Darko Katic, Rainer Jäkel, Michael Beetz, Gisela Lanza and I show how neurosymbolic AI can be used across the entire robot programming and deployment process. BANSAI was accepted to the 2024 CIRP Conference for Manufacturing Systems (CMS). You can find the paper here: https://lnkd.in/emSxWnRn #neurosymbolic #ai #robotics

🏭 🦾 Robotics is revolutionizing the industry, enabling automation, precision, and efficiency in manufacturing processes. Advanced robots, collaborating with humans in workplaces, are changing the way tasks are executed. ⌨ ⚙️ Through programming, machine learning and artificial intelligence, robots are becoming increasingly versatile and adapting to various tasks. Industrial robotics is a step into the future of production.   #Robotics #Industry #MachineLearning #ArtificialIntelligence #AI #BFirstTech #BFT #TechnologiesofSustainableWorld #ESG #ESGFriendly #NonFinancialReporting #ESGReporting #ESGStrategy 

Living in the age of AI and robotics, we are teaching machines what it means to be human—our creativity to AI and our dexterity to robots. In return, we are learning more about ourselves than ever before. In a recent article with Forbes, our Co-Founder and VP of Operations, Stephen McKinley shares his journey from a metalsmithing studio to designing robots, highlighting how integrating human flexibility into machines can revolutionize the industry and our understanding of human abilities. “At the end of the day though, you need to function within constraints, because even people have constraints. And if you want to make something that’s universally useful, it’s not going to be applicable everywhere. So what we’re doing at Ambi is creating machines that can handle any unknown object you might throw at it within the constraints of it being a package.” Read the full article: https://lnkd.in/g4a8YAbi #HandleMore

Researchers at the Technical University of Munich have successfully integrated proprioception into robots using machine learning. This breakthrough enables robots to gain a sense of their body parts and movements, akin to human self-awareness. A significant step towards more efficient and intelligent robotic systems. Dive into the details of this fascinating research: https://lnkd.in/ewtr4xdR #RoboticsInnovation #MachineLearning #TechResearch

📣 Attention Robotics Engineers: Need to know more about developing robot manipulators using digital twins and simulations? Detailed video shows you how to simulate robot algorithms that use AI and automatically generate production code for robot arms and cobots. #robotics #robotarm #cobot #AI #MATLAB https://spr.ly/6042ixR3g

The future of robotics just got a whole lot smarter. Northwestern University's engineers have developed MaxDiff RL, a new AI algorithm that promises to revolutionize the capabilities and safety of robots. By encouraging robots to explore their environments through designed randomness, MaxDiff RL allows for faster and more reliable learning, making robots equipped with this algorithm more adaptable and efficient. The advantages of MaxDiff RL extend beyond just robotics - from self-driving cars to household assistants, this algorithm has the potential to transform various industries. But with increased complexity comes the need for reliable decision-making, making MaxDiff RL a crucial tool for the future of AI and robotics. How can we use this innovative algorithm to enhance the capabilities of robots in everyday scenarios? --- Hi, 👋🏼 my name is Doug, I love AI, and I post content to keep you up to date with the latest AI news. Follow and ♻️ repost to share the information! #aiindustry #smartrobotics #algorithmimprovement

For my final ML project, alongside my team, we developed an integrated machine learning approach for terrain analysis and path optimization in quadruped robots: From K-NN Classification to Q-Learning on Diversified Soil Types. We've combined three areas of machine learning to help robots move across different types of ground. Our recent tests show that the robot is getting better at moving around. We've noticed that it's making fewer unnecessary moves and is more successful in getting around different terrains. The way the robot's path looks and the numbers we see in our data both show that the robot is learning and getting better at handling different ground conditions. I'm proud of our team's innovative approach and look forward to further developments in this space. You can check here the complete models: https://lnkd.in/gSuDnhGR #MachineLearning #Robotics #ArtificialIntelligence #Innovation #TechAdvancements #RoboticNavigation #UnsupervisedLearning #Technology #Engineering #RL #SupervisedLearning

When robots interact with delicate objects, the cost of collisions can be expensive. When those delicate objects are people, the consequences can be dangerous, even fatal. Safeguards exist, but employing them often slows operations to impractical levels. Enter a team of MIT researchers led by MIT Aero/Astro and MechE professor Russ Tedrake. Their algorithms enable millimeter precision in just a few seconds … with 100% accuracy. Not 99%. Not 99.99%. But 100%. Read on here: https://lnkd.in/g3pDgF2Z #mit #manufacturing #robotics #algorithms

Tesla's OPTIMUS ROBOT is learning on the job! Currently, it's being trained by humans to carefully place individual batteries into designated slots. The robot demonstrates impressive fine motor control, but increasing its packing speed is a key next step. video credits : Interesting Engineering #Tesla #Optimus #Robot #AI #Automation #BatteryPacking #Engineering #FutureofWork #Tech #Innovation #Trending #NextLevel #Precision #FineMotorSkills #Robotics #Training #AIRevolution #MachineLearning #SpeedBoost #Efficiency #FutureTech #STEM

Great webinar today from The Robot Report with leaders from Formant discussing AI in robotics: the pragmatic vs. the possible. We are starting to see the robotics industry place bets in AI by increasing funding and spend on machine learning capabilities. Two main areas where AI can make a difference in robotics: robot capabilities (task autonomy/orchestration) & data processing (training data /operations support/ data outside of robotics) Key takeaway - we are 3 to 5 years from deployment of generality robotics fleets aka humanoid robots. The advancement in AI, machine learning and big data allows for more autonomous use in robotics to reduce labor intensive workload and decrease costs to end users. Thank you Steve Crowe for hosting this great webinar and looking forward to the next one!