Memristor-Controlled Robots

Memristor Controlled Robots

The memristor is the recently discovered 4th electronic circuit element of great interest because it operates very similarly to neurons and can be thought of as an artificial synapse.

Memristors are capable of learning and memory, and as such offer a route to more brain-like computation. The non-linearity of the device makes it attractive for complex brain-like control of robotics and the device operates in a very low power way, making memristor-based circuits much more energy efficient than transistor circuitry.

Brain-like computation has been attempted before, but with transistors the final device is much larger than a biological brain. Memristors, being more complex and capable of being fabricated much smaller than transistors, offer an opportunity to create biological-scaled artificial brains with brain-like functionality. As there are theories that suggest that neurons are actually memristive, memristor are the best candidate for this type of unconventional control.

Project Objective

  • Fabricate and characterize memristors
  • Fabricate and characterize memristor circuitry and operations
  • Learning experiments with a memristor controlled robot

Current State of the Project

Memristors have been fabricated and fully characterized. The memristor’s behaviour in circuits has been studied allowing theoretical models of the memristor to be developed. Simulations of memristor-based robot learning in a simple maze navigation task have been performed.

Back to Unconventional Computation

Theme Leader

Project Team

  • Dr Ella Gale
  • Ben de Lacy Costello
    Tel: +44 (0) 117 32 82461
  • Gerard David Howard
  • Larry Bull
    Tel: +44 (0) 117 32 83161

Funding Source

EPSRC Logo
Visit the EPSRC website

Page last updated 12 May 2016

  •