RoboticsPhys.org Technologyabout 1 hour ago
New PRISM Network Improves Robot Control
4 min read
Researchers at University of Michigan Engineering developed PRISM, a neural network component that learns multiplicative interactions for robot motor control. Presented at the 10th Conference on Robot Learning in Austin, Texas in 2026, the system achieved 92.5% success in simulated walking trials versus 51% for conventional networks.
Level
Hype check
Tap to vote and see what everyone thinks.
