Ep#115: Delta-0 (Δ₀): A New Chapter in Humanoid Intelligence
Xiaoen, CEO and founder, and Suen, chief scientist, of Delta Intelligence discuss the Delta-0 humanoid foundation model with RoboPapers hosts Michael and Chris.
Filippo, host of Among Us: Human-Robot Teaming, visits ICRA 2026 in Vienna with researchers and industry leaders to discuss robotics demonstrations, event-camera grasping, markerless motion capture, wearable supernumerary limbs, closed-loop grasping, and European robotics collaboration.
At the exhibition, Filippo tours humanoid and quadruped robots, including Unitree’s G1, and demos such as card playing, popcorn making, and grocery picking. In the handover challenge, teams use cameras and robotic grippers to transfer objects from people to robots, testing whether systems trained on familiar items can handle new ones, including difficult plastic bags. Other challenges involve mobile robot navigation, mazes, and humanoid tasks on rough terrain.
Filippo and Kumail discuss a poster on event-camera recognition and grasp-point detection, co-authored by Hamza Zafar. The model converts event streams into 30-millisecond frames and uses multitask learning; on Event-Grasp it reaches 99% classification accuracy and 93.67% object-detection IoU. Transferring learned layers to the harder NeuroGrasp dataset improves results, and the team says the setup runs locally on a Raspberry Pi without internet access.
Pietro Balatti of the Italian Institute of Technology describes a camera-based motion-capture system that tracks people’s joints and objects without markers or depth cameras. Its camera-agnostic smart-camera nodes can be added to cover larger spaces, and calibration takes under a minute. Applications discussed include safer robot collaboration, workplace ergonomics, rehabilitation, sports movement analysis, and augmented or virtual reality.
MIT researcher Eric Bistolas explains wearable supernumerary robotic limbs, which add appendages rather than supporting or replacing existing limbs. He describes astronaut and Earth applications, including factory work, elder care, and reducing physical strain for military logistics workers. His research explores inferring human intent and predicting movement to make shared control feel more natural; he also discusses high-torque EELS actuators and less invasive haptic feedback.
MIT PhD researcher Kentaro Barahit presents topological loop-closure grasping: a mechanism first reaches around an object with an open tip, then closes into a loop that carries load in tension. The approach aims to combine adaptable grasping with strength for heavy, fragile objects, with examples including patient transfers using a growing vine robot sling. In a later conversation, PAL Robotics CEO and euRobotics president Francesco Ferro discusses interoperability through ROS 2 and collaboration across European research and industry.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.
Xiaoen, CEO and founder, and Suen, chief scientist, of Delta Intelligence discuss the Delta-0 humanoid foundation model with RoboPapers hosts Michael and Chris.
Ted, founder of Prometheus, discusses the verification bottlenecks between AGI and physical abundance at a Humanoid Hub event hosted by Devang.
Do Park of Lightberry explains how the company designs Lumi for emotionally engaging human interaction at The Humanoid Hub.
Aadeel Akhtar of PSYONIC presents at Humanoids Summit on bionic hands that serve prosthetic users and help robots learn manipulation tasks.
Light Origins’ presenter explains how Light-O1 uses human video to train robot action models and demonstrates it on a Lightwheel setup and a Unitree G1.
Joseph Amigo, a second-year PhD student, and Ruhollah, a fourth-year PhD student at NYU, discuss their paper on robot reinforcement learning with RoboPapers hosts.