Software Motion (Suzhou) Engineering Services Co.,Ltd’s Post

It was a great pleasure to speak today at Siemens Digital Industries Software Realize LIVE about AI-accelerated engineering. My talk covered our vision for using AI in engineering processes, focusing on interactive design, generative AI for optimization, and predictive AI alongside simulations. I addressed the accuracy of AI models by comparing AI predictions with simulation results and discussed the key enablers for AI adoption in engineering, including transfer learning and model uncertainty. I had the opportunity to engage in many interesting conversations with users of the technology and the Siemens team. I truly enjoyed the wide range of exciting talks! Thanks to Jean Claude Ercolanelli and Sudhi Uppuluri for the invitation! Feel free to reach out if you’d like to discuss AI in engineering further! #AI #Engineering #AIAutomation #GenerativeAI #PredictiveAI #Siemens #Innovation #Technology #EngineeringExcellence #RealizeLive #MachineLearning #TechTalks #FutureOfEngineering

🔬 Lab Automation: What’s In and What’s Out in 2025 🚀 Our latest blog highlights our view of what's reshaping the future of lab automation this year, such as: ✅ Modular lab systems ✅ Magnetic levitation decks ✅ AI copilots ❌ Microfluidics outside of niche assays ❌ Legacy data tools Discover what's driving innovation, and what’s falling behind in technologies of the Automated Lab in 2025: https://lnkd.in/ghFeTsgp What’s your take? #LabAutomation #LabInnovation #SLAS2025

💥💥💥 Know Where You're Uncertain When Planning with Multimodal Foundation Models: A Formal Framework Neel P. Bhatt, Yunhao Yang, Rohan Siva, Daniel Milan, Ufuk Topcu, Zhangyang Wang Abstract Multimodal foundation models offer a promising framework for robotic perception and planning by processing sensory inputs to generate actionable plans. However, addressing uncertainty in both perception (sensory interpretation) and decision-making (plan generation) remains a critical challenge for ensuring task reliability. We present a comprehensive framework to disentangle, quantify, and mitigate these two forms of uncertainty. We first introduce a framework for uncertainty disentanglement, isolating perception uncertainty arising from limitations in visual understanding and decision uncertainty relating to the robustness of generated plans. To quantify each type of uncertainty, we propose methods tailored to the unique properties of perception and decision-making: we use conformal prediction to calibrate perception uncertainty and introduce Formal-Methods-Driven Prediction (FMDP) to quantify decision uncertainty, leveraging formal verification techniques for theoretical guarantees. Building on this quantification, we implement two targeted intervention mechanisms: an active sensing process that dynamically re-observes high-uncertainty scenes to enhance visual input quality and an automated refinement procedure that fine-tunes the model on high-certainty data, improving its capability to meet task specifications. Empirical validation in real-world and simulated robotic tasks demonstrates that our uncertainty disentanglement framework reduces variability by up to 40% and enhances task success rates by 5% compared to baselines. These improvements are attributed to the combined effect of both interventions and highlight the importance of uncertainty disentanglement which facilitates targeted interventions that enhance the robustness and reliability of autonomous systems. 👉 https://lnkd.in/dH8kw2Vu #machinelearning

If you are interested in building SAFE AI then check out this event by the MathWorks partner, Infineon Technologies. You will be able to hear my co-worker John Kluza present on how you can develop and deploy safe #AI with #MATLAB and #Simulink.

The #V-model from software development has also shaped the #automotive industry for a number of years and has led to greater flexibility and speed. Now it’s time for the next step: accelerated #engineering with #GenerativeArtificialIntelligence. This topic will also be a part of our #AI #TechDay in November, so stay tuned! But until then, here are some of the solutions, which #IAV already presented: • The #RequirementsEngineeringAssistant represents the next generation of requirement analysis. With #REA, requirement lists are not only reviewed, but also optimized – seamlessly integrating into existing systems. This significantly boosts efficiency and precision. • Our #OpenScenarioTool makes the creation of simulation scenarios incredibly simple thanks to AI. With customizable scripts and integration into the IAV #Vega Scenario Simulation Platform, every detail can be considered and perfected. • #TestGPT introduces a breath of fresh air into the generation of test cases for the automotive industry. By combining language technology and requirement analysis, it simplifies and streamlines the process. Curious to learn more? Feel free to reach out to Cindy Xuan, Johannes Dornheim and Michael Reichel. #GenAI #AcceleratedEngineering

𝐄𝐱𝐩𝐥𝐨𝐫𝐢𝐧𝐠 𝐇𝐢𝐞𝐫𝐚𝐫𝐜𝐡𝐢𝐜𝐚𝐥 𝐈𝐧𝐭𝐞𝐫𝐚𝐜𝐭𝐢𝐨𝐧 𝐂𝐨𝐧𝐭𝐫𝐨𝐥 𝐟𝐨𝐫 𝐌𝐨𝐝𝐮𝐥𝐚𝐫 𝐑𝐨𝐛𝐨𝐭 𝐌𝐚𝐧𝐢𝐩𝐮𝐥𝐚𝐭𝐨𝐫 𝐒𝐲𝐬𝐭𝐞𝐦𝐬 I recently delved into hierarchical control strategies for modular robot manipulator systems, inspired by innovative approaches in human-robot collaboration. To deepen my understanding, I developed a MATLAB simulation exploring key concepts and their applications. 𝐊𝐞𝐲 𝐂𝐨𝐧𝐜𝐞𝐩𝐭𝐬 𝐄𝐱𝐩𝐥𝐨𝐫𝐞𝐝: 𝐇𝐮𝐦𝐚𝐧-𝐑𝐨𝐛𝐨𝐭 𝐂𝐨𝐥𝐥𝐚𝐛𝐨𝐫𝐚𝐭𝐢𝐨𝐧 (𝐇𝐑𝐂): A critical focus for advancing modular robotic systems, blending human flexibility with robotic precision. 𝐒𝐭𝐚𝐜𝐤𝐞𝐥𝐛𝐞𝐫𝐠 𝐆𝐚𝐦𝐞 𝐓𝐡𝐞𝐨𝐫𝐲: A hierarchical approach that reflects the leader-follower dynamics between humans and robots. 𝐀𝐝𝐚𝐩𝐭𝐢𝐯𝐞 𝐃𝐲𝐧𝐚𝐦𝐢𝐜 𝐏𝐫𝐨𝐠𝐫𝐚𝐦𝐦𝐢𝐧𝐠 (𝐀𝐃𝐏): Utilized to address Hamilton-Jacobian equations for optimal control. 𝐒𝐢𝐦𝐮𝐥𝐚𝐭𝐢𝐨𝐧 𝐎𝐯𝐞𝐫𝐯𝐢𝐞𝐰: The simulation showcases: 𝟏. 𝐎𝐛𝐬𝐞𝐫𝐯𝐚𝐭𝐢𝐨𝐧 𝐄𝐫𝐫𝐨𝐫𝐬: Tracking the discrepancy between system states and observed data. 𝟐. 𝐓𝐫𝐚𝐜𝐤𝐢𝐧𝐠 𝐏𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞: Demonstrating precise control to meet desired trajectories. 𝟑. 𝐃𝐢𝐬𝐭𝐮𝐫𝐛𝐚𝐧𝐜𝐞 𝐀𝐭𝐭𝐞𝐧𝐮𝐚𝐭𝐢𝐨𝐧: Highlighting the robustness of the system under external disturbances. 𝟒. 𝐎𝐛𝐬𝐞𝐫𝐯𝐞𝐫 𝐚𝐧𝐝 𝐂𝐨𝐧𝐭𝐫𝐨𝐥 𝐈𝐧𝐩𝐮𝐭 𝐏𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞: Visualizing dynamic feedback and the role of neural network weights. 📊 𝐊𝐞𝐲 𝐈𝐧𝐬𝐢𝐠𝐡𝐭𝐬: The results confirmed the effectiveness of hierarchical Stackelberg strategies in achieving uniform boundedness of errors, even under disturbances, as validated by Lyapunov theory. 🎯 𝐓𝐚𝐤𝐞𝐚𝐰𝐚𝐲: Such frameworks open doors to human-centered robotics applications, including medical assistance, rehabilitation, and collaborative manufacturing. 🔗 Want to discuss this further? Let’s connect and collaborate on innovations in control systems and robotics! 𝐒𝐢𝐦𝐮𝐥𝐚𝐭𝐢𝐨𝐧 𝐫𝐞𝐬𝐮𝐥𝐭:

Generative AI is rapidly transforming #automotive software development, offering potential time savings of up to 30 to 50 percent in coding and validation tasks. With the growing complexity of software-defined vehicles, #AI-driven tools are enhancing productivity by automating repetitive tasks, improving code quality, and enabling seamless collaboration between OEMs and suppliers. However, the adoption of GenAI also brings challenges—ensuring data security, regulatory compliance, and seamless integration within existing development pipelines remain critical hurdles for the industry. Striking the right balance between automation and human oversight is essential to maintaining functional safety and cybersecurity standards. McKinsey & Company’s latest insights provide a deep dive into how GenAI is shaping the future of automotive software development. Read more here: https://lnkd.in/gteh7ZST

We would like to thank Altair for supporting our team with its software, helping us accomplish our goals in our recent competition! Altair is a global leader in computational science and artificial intelligence (AI). From groundbreaking simulation technology to cutting-edge AI solutions, Altair Engineering empowers organizations to tackle complex challenges and drive innovation across industries. With a strong focus on high-performance computing and data-driven design, Altair’s tools and expertise help clients unlock powerful insights, optimize performance, and create smarter, more efficient products. Their impact spans aerospace, automotive, healthcare, and beyond, proving that they are not just a technology provider but a true partner in engineering success. Θα θέλαμε να ευχαριστήσουμε την Altair που υποστήριξε την ομάδα μας με τα λογισμικά της, βοηθώντας μας να επιτύχουμε τους στόχους μας στον πρόσφατο διαγωνισμό μας! #AltairEngineering #AI #onlyforward #robotics #engineering #electronics

Monolith, the artificial intelligence (AI) software provider, has been selected by the automotive engineering consultancy HORIBA MIRA as its AI partner for data-driven battery and powertrain development and testing. Read more here: https://lnkd.in/gS3UmAtN #manufacturing #engineering #logistics

Engineering simulation has always been crucial for refining and validating designs. Yet, traditional CAE simulation faces obstacles like hardware constraints and data siloes. 🌐 Simulation-assisted AI transforms automotive design and validation, breaking free from hardware limitations with cloud-native technology. What's more, it promotes teamwork, bridging data siloes and enabling smooth project sharing. 👉 Dive into our latest blog post to uncover the essential approaches of AI simulation and its four key advantages: https://hubs.la/Q02vvLDH0 #automotive #AI #innovation