Robotics Solutions for Cognitive Support

Teaching Robots to Remember: How Cognitive Architectures and Knowledge Graphs Are Solving LLM Limitations in Social Tutoring ... What if robots could adapt to your needs like a skilled tutor – remembering past interactions while balancing conversation and task goals? 👉 WHY THIS MATTERS Today’s AI tutors face a critical challenge: Large Language Models (LLMs) excel at generating human-like text but struggle with three key areas: - Contextual continuity: Maintaining coherent dialogue over extended interactions - Social-task balance: Switching between friendly conversation and instructional focus - Experience integration: Using past interactions to personalize future sessions Traditional systems either rigidly follow scripts or rely on reinforcement learning, which struggles with complex social dynamics. 👉 WHAT THE RESEARCH SOLVES A team at the Italian Institute of Technology developed a cognitive architecture for robots that: 1. Guides yoga sessions autonomously using an LLM agent to choose actions (demonstrate poses, give feedback, chat) 2. Builds persistent memory through knowledge graphs that store user preferences, performance data, and conversation summaries 3. Balances social rapport with task efficiency – no more robots that lecture endlessly or chat aimlessly 👉 HOW IT WORKS The system combines three innovations: 1. Interaction Manager: An LLM agent acts as a "conductor," choosing when to demonstrate poses, ask questions, or provide feedback based on real-time context 2. Working Memory: Stores critical session details (completed poses, user skill level) to handle interruptions naturally 3. Knowledge Graph: Converts raw interaction data into searchable relationships (e.g., "User A struggles with balance poses → suggest modified exercises") Key results from 10-user study: - 100% success in choosing correct interaction steps - 1.1-second average response time (comparable to human tutors) - Users reported the robot felt "attentive but focused" during sessions 👉 Surprising insight: When tested against traditional AI memory systems, the knowledge graph approach matched standard methods on simple queries but outperformed them by 28% on complex questions requiring multi-step reasoning (e.g., "Which poses do beginners find hardest?"). 👉 Why this changes the game: By combining LLMs with structured memory systems, robots gain two human-like capabilities: 1. Building personalized relationships through accumulated interaction history 2. Making context-aware decisions that blend social warmth with task efficiency The architecture’s modular design allows adaptation to other domains like elderly care or collaborative manufacturing. Current work focuses on improving non-verbal cues to make pauses feel intentional rather than system lag.

🧠 Can AI and Robotics Rethink Dementia Care? As the population grows older, millions of individuals face the daily challenges of Alzheimer’s disease and related dementias. Families and caregivers often navigate this journey with limited tools and immense emotional strain. But what if technology could support, rather than replace, the human connections that define compassionate care? Recent research advancements show that AI-enhanced robotics can: ✅ Support cognitive engagement through music and conversation ✅ Learn and adapt to individual routines and emotional cues ✅ Assist caregivers with timely reminders, coaching, and emotional regulation ✅ Integrate with sensors, wearable devices, and electronic health records to provide more comprehensive care Still, the real work lies ahead. We need a national strategy that unites engineering, ethics, and empathy. This includes open access platforms, cross sector collaboration, and inclusive policies that ensure these innovations reach those who need them most. 🧭 The future of dementia care will be collaborative: human and AI, empathy and precision. Let us build systems that not only extend life, but also protect dignity.

𝐄𝐦𝐩𝐨𝐰𝐞𝐫𝐢𝐧𝐠 𝐑𝐞𝐡𝐚𝐛𝐢𝐥𝐢𝐭𝐚𝐭𝐢𝐨𝐧 𝐰𝐢𝐭𝐡 𝐑𝐨𝐛𝐨𝐭𝐢𝐜 ‘𝐂𝐨𝐚𝐜𝐡𝐞𝐬’🤖🧠 Exciting advancements in rehabilitation technology! The National Robotarium's Human-Robot Interaction team has successfully trialed robotic ‘coaches’ in Vienna, Austria, to support stroke and brain injury survivors. These innovative robots, developed in partnership with the AIT Austrian Institute of Technology, interact with patients using a headset that detects brain activity. By understanding the user's intended movements, the robots provide real-time motivation, visual mimicry, and feedback, making rehabilitation more engaging and personalized. Given that 80% of stroke and brain injury survivors experience upper limb impairments, this technology could be a game-changer. Traditional rehab often requires repetitive, task-specific exercises, but with 60% of stroke survivors struggling with forgetfulness and only 31% completing their routines, adherence is a significant challenge. This new approach aims to address that by tailoring support to each patient's needs and cognitive abilities. Over a three-month study, 16 survivors and six therapists tested this system, and the feedback was positive, highlighting the robots' ability to motivate and build trust. Funded by the EU’s Horizon2020 Research and Innovation Programme, this project is a promising step forward in rehabilitation technology, emphasizing the importance of European collaboration in healthcare. Read more and watch a longer video here: https://lnkd.in/eDdeFU3n

Loneliness is one of the biggest health threats seniors face daily. But robotic companions are changing the story... They're bringing connection, comfort, and even safety to senior care. The outcomes? → Less loneliness, more joy. Robotic pets and companions provide emotional comfort, human-like interaction, and a sense of presence (even in the absence of family). → Sharper minds, fewer symptoms. Cognitive games, conversations, and responsive activities help keep seniors mentally engaged. Studies show reduced agitation and depression, especially in those with dementia. → More connection, less isolation. These robots encourage social participation and even spark new friendships in communities. Some help facilitate video calls and group reminders. → Greater safety, more independence. From medication reminders to fall detection and emergency alerts, some robotic companions go far beyond emotional support. We’re not just integrating tech. We’re reimagining companionship. At Eldermark, we believe every tool that supports dignity, connection, and mental health has a place in the future of senior care. And robotic companions are proving they can do all three. Could this be one of the most human uses of AI yet?

Could a robotic puppy become an FDA-registered medical device that can serve in hospitals and other care facilities? That’s what US robotics company Tombot is striving for with Jennie, a highly realistic looking AI-powered robot dog designed to provide companionship and comfort to those struggling with cognitive health challenges like dementia. Tombot CEO Tom Stevens was inspired to develop Jennie because of his experience with his mother's dementia and the challenges of real pet ownership in such situations. The dog is powered by a rechargeable battery, produces AI-generated barks and puppy sounds in response to voice commands and has touch sensors for responsiveness to petting. It also comes with a smartphone app for naming and monitoring engagement. The canine companion is primarily aimed at seniors with dementia or mild cognitive impairment, but could also be useful for individuals with mental health conditions such as depression or PTSD. Simply put: Jennie is an accessible alternative for people who can’t have animals, providing companionship without the challenges of real pet ownership. In 2020, there were over 55 million people worldwide living with dementia. Unless someone finds a cure (quite some companies are investing in this, fortunately), this number is expected to almost double every 20 years, reaching 78 million in 2030 and 139 million in 2050. In a world with an aging population, increased dementia, increased loneliness and difficult access to elder care (especially because of a lack of personnel), robot solutions like these are truly relevant. Maybe AI will be a help in solving one of the biggest issues of our time: loneliness and lack of humans working in healthcare.