Why Robot Programming Adaptability Matters

I watched our humanoid make coffee in the office kitchen. The milk was not where it was yesterday. A mug was half blocked by plates. Nothing was scripted. The robot adapted and kept going. That is the point. Kitchens are messy. Objects move. Layouts change. Interruptions happen. If a robot can operate there without freezing or breaking things, it can handle the edge cases that matter in logistics and industry. Most robotics demos avoid this. Fixed objects. Clean setups. Repeatable motions. That is not the real world. At Sereact we build for live, unscripted environments. The coffee is irrelevant. Learning to generalise is everything. If it works in our kitchen, it works in your warehouse.

𝗬𝗼𝘂𝗿 𝗙𝗮𝗰𝘁𝗼𝗿𝘆 𝗪𝗶𝗹𝗹 𝗥𝗲𝗰𝗼𝗻𝗳𝗶𝗴𝘂𝗿𝗲 𝗜𝘁𝘀𝗲𝗹𝗳. 𝗢𝗿 𝗦𝗼𝗺𝗲𝗼𝗻𝗲 𝗘𝗹𝘀𝗲’𝘀 𝗪𝗶𝗹𝗹 Factories used to be built to last. Now they’re being built to 𝗮𝗱𝗮𝗽𝘁. The next era of manufacturing belongs to 𝗦𝗺𝗮𝗿𝘁 𝗥𝗲𝗰𝗼𝗻𝗳𝗶𝗴𝘂𝗿𝗮𝗯𝗹𝗲 𝗠𝗮𝗻𝘂𝗳𝗮𝗰𝘁𝘂𝗿𝗶𝗻𝗴 𝗦𝘆𝘀𝘁𝗲𝗺𝘀 (𝗦𝗥𝗠𝗦) — machines that don’t wait for a change request. They become the change. What makes them different Instead of pushing data one way — from sensors to screens — SRMS close the loop. When 𝗱𝗲𝗺𝗮𝗻𝗱 𝘀𝗵𝗶𝗳𝘁𝘀 or a 𝗰𝗲𝗹𝗹 𝗳𝗮𝗶𝗹𝘀, the system 𝗿𝗲𝗰𝗼𝗻𝗳𝗶𝗴𝘂𝗿𝗲𝘀 𝗮𝘂𝘁𝗼𝗺𝗮𝘁𝗶𝗰𝗮𝗹𝗹𝘆. Machines redistribute work. Logic updates itself. Production keeps moving. That’s not theory — it’s happening now. At Bosch’s Homburg plant, cells adapt to 200+ product variants daily. At Siemens’ Amberg factory, 99% automation uptime comes from live self-diagnostics. And Fraunhofer’s Plug & Work architecture cuts integration times by up to 70% (Morgan et al., Journal of Manufacturing Systems, 2021). 𝗪𝗵𝘆 𝗶𝘁 𝗺𝗮𝘁𝘁𝗲𝗿𝘀 Because adaptability is the new industrial advantage. Speed of change now beats scale of output. SRMS make it real: • No line rebuilds for new SKUs. • No downtime for reprogramming. • No lag between data and action. They turn factories into dynamic ecosystems — learning, evolving, and responding in real time. 𝗧𝗵𝗲 𝘀𝗵𝗶𝗳𝘁 This isn’t about automation. It’s about 𝗮𝘂𝘁𝗼𝗻𝗼𝗺𝘆. The smartest factories aren’t the biggest. They're the ones that retool themselves mid-shift —  before the market even moves. The new industrial power isn’t in mass production. It’s in 𝗺𝗮𝘀𝘀 𝗮𝗱𝗮𝗽𝘁𝗮𝘁𝗶𝗼𝗻. And SRMS are how you build for that. Ref: Journal of Manufacturing Systems- Vol 59

Many enterprises are still experimenting with 𝐀𝐈 𝐚𝐠𝐞𝐧𝐭𝐬 as if they are just “better chatbots.”⠀ That view is already outdated.⠀ ⠀ A new survey on 𝐀𝐠𝐞𝐧𝐭𝐢𝐜 𝐑𝐞𝐢𝐧𝐟𝐨𝐫𝐜𝐞𝐦𝐞𝐧𝐭 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 (𝐀𝐠𝐞𝐧𝐭𝐢𝐜 𝐑𝐋) reframes the conversation. It positions LLMs not as static generators of text, but as autonomous decision-makers capable of planning, reasoning, using tools, remembering, learning from feedback, and adapting in dynamic environments. ⠀ For business leaders, this shift matters.⠀ Why? Because enterprise use cases rarely live in one-shot Q&A. They live in workflows, processes, and environments where context changes constantly.⠀ ⠀ Key takeaways that executives should note:⠀ ⠀ 𝐅𝐫𝐨𝐦 𝐒𝐭𝐚𝐭𝐢𝐜 𝐭𝐨 𝐀𝐝𝐚𝐩𝐭𝐢𝐯𝐞: Traditional reinforcement learning for LLMs (RLHF, DPO) optimizes single-turn responses. Agentic RL trains systems to act across multiple steps, adjusting to feedback and uncertainty - closer to how real business operates.⠀ ⠀ 𝐄𝐧𝐭𝐞𝐫𝐩𝐫𝐢𝐬𝐞-𝐑𝐞𝐚𝐝𝐲 𝐂𝐚𝐩𝐚𝐛𝐢𝐥𝐢𝐭𝐢𝐞𝐬: Planning, memory, tool-use, and self-improvement are not research curiosities. They map directly to enterprise needs like automated research, iterative code generation, financial analysis, and customer service orchestration.⠀ ⠀ 𝐄𝐧𝐯𝐢𝐫𝐨𝐧𝐦𝐞𝐧𝐭𝐬 𝐌𝐚𝐭𝐭𝐞𝐫: Just as self-driving cars need simulated roads, enterprise AI agents need domain-specific environments - ERP, CRM, supply chain platforms - where they can train, fail safely, and improve.⠀ ⠀ 𝐎𝐩𝐞𝐧 𝐂𝐡𝐚𝐥𝐥𝐞𝐧𝐠𝐞𝐬: Trustworthiness, scalability of training, and robustness of environments are highlighted as unresolved. For enterprises, this means governance, safe deployment sandboxes, and evaluation frameworks will be critical.⠀ ⠀ 𝐓𝐡𝐞 𝐛𝐨𝐭𝐭𝐨𝐦 𝐥𝐢𝐧𝐞:⠀ Agentic RL marks the next phase of enterprise AI. The ROI will not come from deploying a model once, but from building systems that learn, adapt, and improve inside your business workflows.⠀ ⠀ The companies that invest in building these environments and training loops - just as they once invested in data lakes and DevOps pipelines - will define the next generation of enterprise advantage.

Physical AI: Breaking the Boundaries of Robotics The real revolution in robotics isn’t just mobility - it’s adaptability. We’re entering an era where AI and robotics are fusing into Physical AI, unlocking new learning strategies that transcend fixed designs and pre-programmed motions. This means machines can: ✅ Learn from experience, not just follow instructions ✅ Adapt to any shape and function, from humanoids to bio-inspired designs ✅ Overcome physical constraints, operating in environments previously out of reach Unlike traditional automation, where robots are built for narrow use cases, Physical AI enables systems to reshape, reconfigure, and respond dynamically to real-world challenges. Machines that don’t just move but continuously evolve - optimizing performance through reinforcement learning and real-time physics adaptation. This shift isn’t just about better robots - it’s about entirely new categories of machines that can thrive in unstructured, unpredictable environments. From soft robotics in medicine to autonomous systems navigating extreme terrains, Physical AI is expanding the limits of what machines can do. What’s next? Could we see AI-powered systems that dynamically reassemble, shift between locomotion styles, or autonomously repair themselves?

Why do powerful pretrained generalist robot models fail when you move an object a few inches, swap a target, or change the scene layout? It’s usually not a lack of motor skill — it’s an alignment problem at test time. In our new paper, we introduce Vision–Language Steering (VLS): a training-free, inference-time framework that adapts frozen diffusion and flow-matching robot policies to out-of-distribution (OOD) scenarios. Key idea: Treat adaptation as an inference-time control problem. Instead of retraining policies, we steer the denoising process using: -Vision–Language Models to interpret test-time constraints -Differentiable, programmatic rewards grounded in 3D geometry -Gradient-based guidance + particle resampling for stable long-horizon execution 📊 Results CALVIN: +31% absolute success over prior steering methods LIBERO-PRO: +13% improvement on strong VLAs (π0.5, OpenVLA) Real world (Franka): Robust execution under appearance shifts, position swaps, and novel object substitutions This work suggests a broader takeaway for robotics foundation models: Scaling policies alone isn’t enough — inference-time alignment matters. 📄Paper: https://lnkd.in/g67pf5Tm 🌐 Project page: https://lnkd.in/gkPxZjXw

Robotics is entering a new phase where learning is becoming more autonomous, scalable, and efficient. Instead of relying heavily on large volumes of human-labeled training data, emerging approaches allow robots to learn through simulation, self-exploration, and real-time adaptation. This shift has the potential to significantly reduce development time while improving flexibility across dynamic environments. In practical terms, this means robots can better understand how to interact with unfamiliar objects, refine their movements through trial and feedback, and generalize skills across tasks without being explicitly programmed for each scenario. From manufacturing floors to logistics and even healthcare support, the impact could be substantial. While the progress is promising, it also brings important considerations around reliability, safety, and oversight. As robots gain more independence in how they learn and act, ensuring robust validation and responsible deployment becomes critical. The evolution from data-dependent training to self-directed learning is not just a technical milestone. It represents a broader shift toward more adaptive and intelligent systems that can collaborate with humans more effectively and operate in increasingly complex real-world settings.

The Future of Robotics Isn’t Just Smarter Machines, It’s Machines That Learn Like HUMANS A breakthrough in reinforcement learning (RL) is quietly rewriting the rules of robotics. Forget rigid, pre-programmed bots—GRPO (Group Relative Policy Optimization) is enabling robots to adapt, compare, and improve like humans. But scaling this tech is harder than it looks. Let’s break it down: Why Traditional Robotics Is Hitting a Wall. Most robots today rely on fixed reward systems: “Pick up cup = +1 point” “Drop cup = -1 point” This works for simple tasks but crumbles in dynamic environments (e.g., handling irregular objects, adapting to human interruptions). GRPO flips the script: Evaluates groups of actions and assigns relative rewards (e.g., “Grip A outperformed Grip B”). Eliminates need for complex value models—cuts compute/memory costs by ~50%. Enables human-like trial-and-error learning through synthetic data. Synthetic Data, The Unsung Hero - Tools like NVIDIA Isaac Sim and DeepSeek’s synthetic engines let robots train 24/7 in hyper-realistic simulations: Autonomous vehicles practice navigating flooded roads. Surgical bots master sutures on virtual patients. Industrial arms adapt to chaotic assembly lines. No real-world risks. No privacy concerns. Just scalable, ethical training. The Roadblocks (and Why They Matter) GRPO isn’t plug-and-play for robotics yet: Sim-to-real gaps: Physics in simulations ≠ real-world friction/noise. Action complexity: Robots deal with continuous movements (e.g., joint angles), not discrete tokens. Compute hunger: Training requires serious GPU firepower (looking at you, NVIDIA L40S). But teams like DeepSeek and Field AI are already showing 5-13% ROI gains in early trials. What This Means for AI Developers Robots trained with GRPO + synthetic data could: Autonomously adapt to factory floor changes. Refine surgical techniques through 10,000 simulated ops. Navigate crowded spaces using “experience” from synthetic NYC sidewalks. The future isn’t just automation—it’s robots that learn on the job. Are you building the next gen of adaptive robots?

Yesterday, I was doing a 5-minute voice chat with a voice agent. I asked a question. The agent replied. I repeated myself & tried explaining what I actually needed. The agent replied again. Same line. So I raised my tone "Are you not getting it? This is my exact requirement." The agent still replied… with same words & same tone... At that point, it didn’t matter how accurate the answer was. It felt like talking to a wall. Most AI agents today are built to respond. But support isn’t just about responding. It’s about reading the room. It's about understanding the "unsaid". When someone is confused, frustrated, or clearly not being understood, the agent needs to shift the tone, speed, what it says and how it says it. That’s the missing loop in most systems today: "Adaptability". Tools like Deepgram can help agents hear what’s beneath the words which frustration, interruptions, hesitations. Resemble AI gives control over how the reply sounds. It can be less chirpy, more grounded. Hey Dev mirrors that visually & let the face react with more awareness. The real challenge? Building systems that understand "human escalation" and adjust accordingly. If an AI agent that can’t shift in moments like this isn’t a support rep. It’s a loud FAQ with a smile. And in a world of AI, being human might just be the ultimate competitive advantage.

🔴 NEW ARTICLE: "VERSES AI Leads Active Inference Breakthrough in Robotics." My latest article breaks down VERSES' newest research paper titled, “Mobile Manipulation with Active Inference for Long-Horizon Rearrangement Tasks,” that was oh-so-quietly released to the public a few weeks ago (Shhh 🤫 ) This new research, led by Dr. Karl Friston's team at VERSES is the blueprint for a new robotics control stack that achieves an inner-reasoning architecture comprised of a hierarchy of multiple active inference agents within a single robot body, all working together for whole-body control to adapt and learn from moment to moment in unfamiliar environments without any offline training. ◼️ Key Takeaways: Instead of a single, monolithic Reinforcement Learning (RL) policy, their architecture creates a hierarchy of intelligent agents inside the robot, each running on the principles of Active Inference and the Free Energy Principle, outperforming current robotic paradigms on efficiency, adaptability, and safety - without the data and maintenance burden of reinforcement learning. Here’s what’s different: 🔸 Agents at Every Scale - Every joint in the robot’s body has its own “local” agent, capable of reasoning and adapting in real time. These feed into limb-level agents (e.g., arm, gripper, mobile base), which in turn feed into a whole-body agent that coordinates movement. Above that sits a high-level planner that sequences multi-step tasks. 🔸 Real-Time Adaptation - If one joint experiences unexpected resistance, the local agent adjusts instantly, while the limb-level and whole-body agents adapt the rest of the motion seamlessly — without halting the task. 🔸 Skill Composition - The robot can combine previously learned skills in new ways, enabling it to improvise when faced with novel tasks or environments. 🔸 Built-In Uncertainty Tracking - Active Inference agents model what they don’t know, enabling safer, more cautious behavior in unfamiliar situations. The result: a robot that can walk into an environment it has never seen before, understand the task, and execute it — adapting continuously as conditions change. VERSES’ broader research stack ties this directly into scalable, networked intelligence with AXIOM, Variational Bayes Gaussian Splatting (VBGS), and the Spatial Web Protocol. Together, these form the technical bridge from a single robot as a teammate to globally networked, distributed intelligent systems, where every human, robot, and system can collaborate through a shared understanding of the world. The levels of interoperability, optimization, cooperation, and co-regulation are unprecedented and staggering. Every industry will be touched by this technology. Smart cities all over the globe will come to life through this technology. ➡️ Get the full story here: 🔗 https://lnkd.in/ghFizkhn #ActiveInferenceAI #AXIOM #VBGS #Robotics #VERSESAI

What happens when a robot loses a leg mid-mission? Most robots would fail immediately. But watch this one figure out how to walk again in just a few tries. The researcher deliberately damages the robot. Cuts off a leg. Adds weights. Attaches wheels to limbs. Each time, the robot experiments with different gaits until it finds one that works. This is omni-bodied intelligence. The software doesn't panic when the hardware changes. It adapts. Here's why this matters: we talk about robots in homes and factories, but we rarely talk about what happens after six months of use. Parts break. Joints wear out. Sensors fail. If robots can't handle imperfection, they'll never leave the lab. This approach treats adaptability as a core feature, not an edge case. That's the difference between a demo and a tool you can actually rely on. Video credits: SkildAI --- Interested in starting your robotics career? Check out our free robotics career guide to get you started: https://lnkd.in/gpPVTPKE