FEATURED ROBOTICS
Underwater Robots Learn From Experience for Safer Navigation
From arXiv • Latest Research
Scientists developed a new system to help underwater robots navigate complex ocean environments more efficiently and safely. This system uses data from sensors and real-time conditions to make better decisions about how to move, which can improve the robot's performance and keep it safe in unpredictable situations.
Read Full Paper →
Artificial Intelligence Learns from Data to Predict User Preferences
Scientists created a new way to imagine and understand the world, by developing a virtual "world model" that can help artificial intelligence systems make decisions. This breakthrough could lead to more advanced and intelligent virtual agents, which could have significant benefits in fields like science, entertainment, and education.
Read Paper →
AI Learns to Shape Its Own Future
Scientists developed a new algorithm called MorphIt that helps robots adapt their shape to suit different tasks. This could lead to more capable and efficient robots that can navigate complex spaces and interact with objects more accurately, by changing their physical form on the fly.
Read Paper →
AI Robot Gets Better at Reading Your Preferences by Analyzing More Data
The researchers have developed a robot that can move around a space facility to inspect and assess its condition, as well as the components stored there. This robot, called the Mobile Inspection Module, has several sensors and tools on board that allow it to take pictures, measure surfaces, and even perform maintenance tasks - all while working independently in space.
Read Paper →
Smarter Robots with Less Data Requirements Robots Now Understand Visual Cues Efficient Vision-Language Models Boost Robotics Capabilities AI for Smaller Devices Edge VLA Puts Robots on the Fast Track AI-Fueled Robots Get Smarter, Not Bigger
Scientists created a new way to make "smart" robots that can understand what they see, called Edge VLA. This innovation could greatly benefit robotics by allowing devices with limited resources to quickly and efficiently understand their surroundings, making it easier for them to perform tasks without getting stuck or needing constant updates.
Read Paper →