Autonomous robotic operation controlled by wave-based neuromorphic hardware
A small robotic car approaches an obstacle. Its distance readings travel by Bluetooth to a dish of water, where six actuators set waves in motion. A camera captures the interference pattern, a classifier reads it and the car receives an instruction to turn. Jacob Zohar and colleagues used this arrangement to demonstrate that a physical wave field can participate in the control of a moving vehicle. Their experiment raises a question that matters whenever a machine acts in the world. Could an independent observer explain why it turned at a particular moment? The published route shows what the car did, and the archived five-class examples show which obstacle configurations that separate classification test included and which it left out. Reconstructing an individual decision requires the readings, images and motor command from the same moment. The distinction also matters for the proposed magnetic successor to the water reservoir, whose obstacle recognition the authors studied through micromagnetic simulation.
Autonomous robotic operation controlled by wave-based neuromorphic hardware Leer más »








