System integration of neuromorphic visual (DVS), processing (SpiNNaker)
and servomotor modules into an autonomous robot controlled by a Spiking
Neural Network
Abstract
This paper describes the design and modus of operation of a neuromorphic
robotic system based on SpiNNaker platform, and its implementation on
the goalkeeper task. The robotic system utilizes an Address Event
Representation (AER) type of camera (DVS) to capture features of a
moving ball, and a servo motor to position the goalkeeper to intercept
the incoming ball. At the backbone of the system is a microprocessor
(Arduino Due) which facilitates the communication between different
robot parts. A spiking neural network, which is running on SpiNNaker,
predicts the location of arrival of the moving ball and decides where to
place the goalkeeper. In our setup, the maximum data transmission speed
of the closed-loop system is approximately 3000 packets per second for
both uplink and downlink, and the robot can intercept balls whose speed
is up to 1 m/s starting from the distance of about 0.8 m. The
interception accuracy is 85%, the response latency is 6.5 ms and the
maximum power consumption is 7.15 W. A demo video of the robot goalie is
available on YouTube: https://www.youtube.com/watch?v=135fH21QXtg
.