Phys.org Robotics News
Daily updates on robotics research breakthroughs and applied technology advances
Unpredictable Movements of Autonomous Robots Can Increase Human Discomfort
Researchers at Toyohashi University of Technology discovered that autonomous robots moving unpredictably trigger sustained human discomfort, as measured by elevated arousal and skin‑conductance responses in a VR passing scenario. Predictable straight‑line motion caused initial arousal that habituated over repeated encounters, while stop‑and‑go behavior kept arousal high. The study, published in the International Journal of Social Robotics, underscores predictability as a key design factor for comfortable human‑robot coexistence. Findings aim to inform motion‑planning algorithms for service robots entering everyday spaces.
New RoboReward Dataset and Models Automate Robotic Training and Evaluation
Stanford and UC Berkeley researchers released RoboReward, a new dataset and benchmark for training vision‑language reward models in robotics. The dataset pairs real‑world robot videos with textual descriptions and graded success scores, enabling models to evaluate task performance automatically. Using...
Adaptive Motion System Helps Robots Achieve Human-Like Dexterity with Minimal Data
Researchers at Keio University have introduced an adaptive motion reproduction system that uses Gaussian process regression to translate human grasping motions to robots with minimal training data. The model learns the relationship between object stiffness and required force/position commands, achieving...
Humanoid Robots or Human Connection? What Elon Musk's Optimus Reveals About Our AI Ambitions
Tesla’s Optimus humanoid robot is positioned as a versatile assistant for factories and homes, with Musk pledging a million units within ten years. The design leverages recent advances in generative AI to give the robot conversational abilities that feel more...