Samir Menon

We are pleased to announce that the article titled "Safe Time-Varying Nonlinear Programming With Equality Constraints", Karthik Shenoy, Arun Mahindrakar, and Umesh Vaidya has been accepted for publication in IEEE Control Systems Letters (L-CSS). The pose estimation of a rigid body in space is a relevant problem that shows up in domains like satellite docking and computer vision for mobile robots. The pose estimation can be reformulated as a time-varying nonlinear programming problem on submanifolds of the Euclidean space (defined using equality constraints), commonly referred to as the time-varying Procrustes problem. This letter focuses on a safety-critical solution to equality-constrained nonlinear programming, where the cost and the constraints vary continuously over time. To address this problem, we propose a continuous-time dynamical system with a quadratic-program (QP)-based feedback. We show that the tracking error dynamics is locally uniformly ultimately bounded, and establish uniform exponential stability in the presence of feedforward prediction. Early access link: https://lnkd.in/dY6DBq2K Pose estimation: https://lnkd.in/dr-9h_rx Control Lab | Indian Institute of Technology, Madras #research #controlsystems #optimization

64

1 Comment

We’re excited to partner with Institute Incubation and Innovation Council (I3C), IIT Jammu to launch certification programs in the areas of Automotive Design & Analysis, Electric Vehicles & Autonomous Vehicles. In the last 3 years, Skill-Lync worked with 60+ Automotive OEM/Tier-1/GCCs and ER&D Services Companies to upskill their workforce on the latest Automotive technologies. We are bringing all of this knowledge together in the form of these certification programs. This is another step toward bridging the long-standing gap between academic learning and industry expectations. We want learners to benefit from the combined strengths of both ecosystems. MT Arvind Dr. Lakhvinder Singh Sarangarajan V Iyengar SuryaNarayanan PaneerSelvam Dhruv Mathur Priyadarshini V Vatsal Shah Vinay KB Shilpa Murali Subramanian J H Tarun Tewari Anand Prasanna Iron Pillar #Skills #Automotive #ElectricVehicles #IITJammu #SkillLync

345

12 Comments

🎉 Excited to share that our research paper on assistive robotics has been published on arXiv! 📄 **"Advancing Assistive Robotics: Multi-Modal Navigation and Biophysical Monitoring for Next-Generation Wheelchairs"** 👨‍🔬 Authors: Md. Anowar Hossain, Mohd. Ehsanul Hoque 🔗 Read the full paper: https://lnkd.in/gVtmYe35 Our work addresses a critical challenge in assistive technology: providing people with mobility impairments greater independence while ensuring their safety and health are continuously monitored. **Key Highlights:** 🔹 **Multi-modal control system** integrating four complementary interfaces: joystick, speech recognition, hand gesture, and electrooculography (EOG) - allowing users to choose the most comfortable control method 🔹 **Real-time health monitoring** tracking heart rate variability, SpO2, and skin temperature with high accuracy (≤2 bpm for heart rate, ≤0.5°C for temperature, ≤1% for SpO2) 🔹 **Impressive performance** 🔹 **Cloud-enabled caregiver support** providing real-time alerts through encrypted data transmission This research contributes toward compliance with ISO 7176-31 and IEC 80601-2-78 safety standards and establishes a foundation for future adaptive machine learning enhancements in assistive robotics. #AssistiveTechnology #Robotics #Healthcare #Innovation #Accessibility #MachineLearning #Research #MobilityAids #HealthTech

7

Interviewed a Final-Year IIT-D Student. What He Said About Student Ideas Shattered Everything I Thought I Knew... ​ As founder of Zentro Suite Corporation(live at JSS STU, scaling to Tier-1 Universities/Fortune 500), I sat down with a final-year IIT student grinding placements, LeetCode, and campus pains. His weekday? IG scrolls, HN sparks, ideas dying in Notes. Friction? "Cognitive activation energy" + exposure fear. Hooks? Asymmetric progress ("upvotes while I slept") + scarce signals ("1 alum shortlisted"). We dissected DAU flywheels, collab hell, and what turns messy sparks into resume gravity. No fluff—just raw Tier-1 habits. This convo is reshaping our Upcoming community dashboard for 5min morning bursts → evening polish → certifiable outcomes. Reader's Takeaway? University innovation heads: Your students' intelligence is leaking daily. Investors: Edtech DAU isn't gamification—it's identity compounding. Students: What if your Notes became public proof? Guess What!! We already started to build it...!!😉 #EdTech #StudentEngagement #InnovationManagement #IIT #UniversityTech

25

2 Comments

Generative diffusion models for proteins are generally used in most applications with “conditioning”, for example we want the generated model to contain a known binding motif or a catalytic constellation of individual amino acids acids. In practice, this means we give the model a desired sub-structure (say a helical binding fragment, or a “theozyme”) and ask it to build a scaffold around this sub-structure. This works only sometimes and the incorporated motif can be far from our desired geometry. This often means lots of sampling and post-process filtering for getting desired designs, therefore reducing efficiencies and increasing computational cost. Could we do it differently? Could we have guarantees that the desired physical features will be present in the generated structures? Here’s our first attempt applying an elegant approach Nando and team have developed. Ultimately motifs are an approximation or a sample from the underlying space of desired physical interaction(“interaction constraint”) e.g., “I want these hydrogen bonding groups on the target to be engaged”, or a negative interaction constraint “I want an aperture/opening/cavity in my binder so substrate entry and product egress can happen”). This framework allows constraining the diffusion model to a physically relevant region of its phase space. Experimental validation of generated designs is in the works in our lab, and we’re excited to continue incorporating various types of physics-guided interaction constraints in generated proteins! #proteindesign #rfdiffusion #functiondesign

38

Delighted to be part of this new lab working towards reliable autonomous agents using #LeanProver for verification. My thoughts on what we are building: Generative AI is famous for being both powerful and unreliable, though more recently it has become more reliable even as it has grown in power. With the growth of agentic systems the power and usefulness of AI has grown greatly, even as reliability has become critical. A natural way to attempt to overcome issues with reliability are by using a Formal Proof System such as the Lean Prover to verify the correctness of outputs of AI. Our lab will work on bridging autonomous agents with Lean to build reliable AI. This will involve building and extending technology across all domains as well as customising to specific domains that matter. The experience and insights of Emergence in agentic AI will greatly help with this. By building in Bangalore with its strong software industry and with strong connections to the Indian Institute of Science and other academic institutes, we hope that a strong ecosystem in this emerging area takes root in India. Lean FRO Leonardo de Moura Indian Institute of Science (IISc) #LeanLang

103

6 Comments

Transient electronics—comprising degradable devices that disintegrate and disappear after their operational lifetime—have garnered significant attention in recent years due to growing concerns over electronic waste (e-waste). However, the mere degradability or biodegradability of electronic components is not sufficient to guarantee environmental safety. A comprehensive assessment of the degradation by-products is also critical for evaluating the true ecological impact of these devices. Findings reported in this paper reveal that, although the materials used in these devices are inherently degradable, their breakdown can lead to the formation of toxic or environmentally persistent substances. These results underscore the critical need for careful material selection and reevaluation of current design strategies in transient electronics. Beyond degradability, the safety and environmental compatibility of degradation products must be key considerations to effectively address the end-of-life challenges associated with electronic devices. Ravinder S. Dahiya Nature Portfolio Springer Nature #sensors #electronics #advancedmaterials #degradation #sustainability #semiconductor #transient #circular #printedelectronics

32