Robotics

Today's Stories

1. 1

   University students create AI robot that learns to play air hockey by practicing in virtual reality first

   Engineering students at University of British Columbia built a robotic air hockey table that trained an AI player in computer simulation, then transferred those skills directly to the physical robot. The project combined computer vision (AI that can see and understand images), reinforcement learning (AI that improves through trial and error), and robotics hardware over two years.

   This shows how robots can now learn complex physical tasks in virtual environments before touching real equipment, making robot training faster and safer for real-world applications.

2. 2

   NVIDIA launches Cosmos 3 AI system that can generate videos and control robot actions from text descriptions

   NVIDIA released Cosmos 3, an AI system that can create realistic videos, images, and robot control commands from written instructions. The system comes in two sizes - a 16 billion parameter version and a larger 64 billion parameter model designed for building AI applications that interact with the physical world.

   This technology could eventually help robots understand complex verbal instructions and practice tasks in realistic simulations before performing them in homes or workplaces.

3. 3

   Researchers develop robotic arm prototype that learns from its own failures to improve grasping objects

   Scientists created a robotic arm system that can select objects to grab, attempt to pick them up, analyze why it failed, and use that failure data to improve future attempts. The system tests different grasping strategies in simulation and maintains safety boundaries to prevent dangerous actions.

   Future household and warehouse robots could become much more reliable at handling unfamiliar objects by learning from their mistakes rather than needing to be programmed for every possible situation.

4. 4

   BlackBerry study finds software problems are the biggest obstacle blocking robotics innovation

   A survey of 1,000 robotics developers worldwide by BlackBerry QNX revealed that software issues, not hardware limitations, are the primary barrier preventing advances in robotics. The study examined how robotics development is changing as systems become more AI-enabled and work alongside humans.

   This suggests that solving software challenges could accelerate the arrival of practical robots in workplaces and homes, since the physical robot hardware is already capable enough.

5. 5

   Uber partners with AI company Autobrains to launch self-driving taxi service in Munich

   Ride-sharing company Uber announced a strategic partnership with Israeli AI firm Autobrains to introduce robotaxi services in Munich, Germany. The announcement caused Uber's stock price to jump 5.2% as investors see potential for autonomous vehicle expansion.

   Customers in Munich may soon be able to request self-driving cars through the Uber app, marking another step toward mainstream autonomous taxi services in major cities.