GeNN is an open source library for generating code to accelerate neural network simulations on NVIDIA GPUs. As these are becoming commonplace, the first aim of this workgroup is to help any interested users get up and running with GeNN and use it to accelerate the simulation of their own models.
The researchers from the University of Sussex at Capocaccia this year are part of a new EPSRC-funded project called Brains-on-Board. Within this project we intend to use GeNN to simulate insect-inspired models of vision, navigation etc that can be run in realtime on robotic platforms powered by hardware such as the NVIDIA Jetson TX1. We have several
simple tracked robots and TX1 modules with us here at Capocaccia and believe that they represent a very promising and flexible platform for embodied robotics research. Therefore, the second aim of this workgroup is to further develop the platform - integrating it with neuromorphic sensors such as the DVS and, inspired by the matching 'Insect brains
as inspiration for neuromorphic systems' discussion group, attempting to use it to solve real world cognitive tasks.
One possible task is a real world Iris classifier where a wheeled robot approaches poor, unsuspecting examples of Sardinian flora and attempts to classify them using input from a camera/DVS.
Login to become a member send
No timetable published yet.
GeNN
GeNN is an open source library for generating code to accelerate neural network simulations on NVIDIA GPUs. As these are becoming commonplace, the first aim of this workgroup is to help any interested users get up and running with GeNN and use it to accelerate the simulation of their own models.
GeNN: a code generation framework for accelerated brain simulations. (1.5 MB)
We will be using the version of GeNN from the development branch on our github also available to download here:
GeNN development (946.5 KB)
Auto-generated HTML documentation including tutorials etc can be downloaded here:
GeNN documentation (3.7 MB)
GeNN can be used natively on Windows, Mac and Linux. However if you want to use Windows, you need to have Visual C++ 2013 or newer installed.
Visually-guided behaviour in insects
Insect-Inspired Visual Systems and Visually Guided Behavior (203.6 KB)
Hardware
We will have two of our prototype NVIDIA Jetson-based autononous robot platforms with us at CapoCaccia. These are installed with GeNN; and GPU accelerated version of OpenCV, VisionWorks and Caffe which should provide a platform capable of running a wide variety of models.
Robot platform
NVIDIA Jetson TX1
NVIDIA Jetson
Jetson TX1
Orbitty carrier board
Orbitty Carrier for NVIDIA® Jetson™ TX2 & Jetson™ TX1
Orbitty Carrier for Jetson
Surveyor Rover
Surveyor Rover MK1
Surveyor Rover