Long-horizon manipulation has always depended on temporal abstraction and semantic reward alignment, and SARM’s stage-aware approach bridges that gap elegantly. At Cybernetic, we’ve observed similar challenges in plug-and-play intelligence, helping robots understand progression within a task, not just the end goal. Incorporating stage-aware reward modeling into shared world models could make real-world generalization far more dependable. Excited to see how frameworks like SARM evolve and integrate with open embodied data pipelines, Qianzhong Chen

Absolutely Incredible phase diagram of the Zeta function on the complex plane. I am working on a "L function automaton" and started with making the complex plane interactive with live scaling of the function. This popped out when I was testing the Zeta function with the system. The critical strip is blazing on the 1/2 real and emergent "crosses" bleeding through the Incredible complexity on the left half. You are looking at the magic behind the prime numbers in a way that I havent seen documented to this exact level of detail. Here is the colab where you can do exactly this. https://lnkd.in/ePC86j4V #Mathematics #VisualMath #ZetaFunction #Lfunction #ComplexAnalysis

Absolutely Incredible phase diagram of the Zeta function on the complex plane. I am working on a L function automaton and started with making the complex plane interactive with live scaling of the function. This popped out when I was testing the Zeta function with the system. The critical strip is blazing on the 1/2 real and emergent "crosses" bleeding through the Incredible complexity on the left half. You are looking at the magic behind the prime numbers in a way that I havent seen documented to this exact level of detail. Here is the colab. https://lnkd.in/ePC86j4V #Mathematics #VisualMath #ZetaFunction #Lfunction #ComplexAnalysis

using UnityEngine; public class RobotController : MonoBehaviour { public Rigidbody rb; public float Kp = 50f, Ki = 0f, Kd = 2f; private float lastError = 0f, integral = 0f; public float speed = 5f; public float turnSpeed = 50f; void FixedUpdate() { // Simplified tilt angle float tilt = transform.eulerAngles.x; if (tilt > 180) tilt -= 360; float dt = Time.fixedDeltaTime; float error = tilt; integral += error * dt; float derivative = (error - lastError) / dt; float pidOutput = Kp * error + Ki * integral + Kd * derivative; lastError = error; Vector3 forwardForce = transform.forward * (speed - pidOutput); rb.AddForce(forwardForce, ForceMode.Force); } public void Turn(float turnInput) { rb.AddTorque(Vector3.up * turnInput * turnSpeed); } public void Stop() { rb.velocity = Vector3.zero; rb.angularVelocity = Vector3.zero; } }

🤖 Modeling Attacks on AI-Powered Apps with the 𝐀𝐈 𝐊𝐢𝐥𝐥 𝐂𝐡𝐚𝐢𝐧 𝐅𝐫𝐚𝐦𝐞𝐰𝐨𝐫𝐤 New framework by NVIDIA. Five stages: recon, poison, hijack, persist, and impact. With an iterate/pivot loop for agentic systems. Great post by Rich Harang describing attacks and defensive priorities for each stage, and how a kill chain might look using a RAG application example, including how defenders can break the attack chain at each stage. https://lnkd.in/gSz_dirz

On this day in tech history: In 1971, the Stanford "Hand–Eye" project quietly produced one of the first systems to integrate vision, reasoning, and robotic manipulation. Its planner operated on symbolic scene descriptions parsed from noisy camera input. Most of the logs weren’t published until internal notes surfaced years later.

Think you need a function to escape nested loops? Wrong. Use labeled statements: `myLoop: for (...) { for (...) { break myLoop; } }`. It's a surgical jump out of deep logic. But it's also a fast track to spaghetti code your team will hate. Is this a forgotten gem for clean control flow, or a `goto` in disguise that should stay buried? #GaboTips

I've spent the afternoon writing an AI agent to diagnose car problems by finding the likely subsystem and component at fault. It's working against a local installation of Llama3 running on Ollama. I'll be talking more about it at the Agents Superstream in a couple of weeks.

We’re building real-time speech-to-speech, which will be the default UI to the humanoid robot F.03 has a 4x more powerful speaker with improved microphone for performance and clarity

The robots are here It’s surreal: a humanoid receptionist by Figure is now being showcased for hotels—meaning traditional front-desk jobs are very much on notice. If your role doesn’t involve management or technical skills, it’s time to upskill. Entry-level jobs as we’ve known them are being redefined. The wave of humanoids working hospitality, manufacturing and logistics is growing fast. Let’s make sure we’re ready.