Agents on the Neurorobotics Platform

I was hoping to get my CABot3 agents running using the virtual environment from the Neurorobotics Platform.
So, I contacted the NRP folks (who have been very helpful).
After a virtual conference, we agreed that the best thing would be to download their software, and run a local server.
After a few weeks waiting to get approved (Andrew, can that process be streamlined for people who already have approval to use one HBP platform?) I got access to their software.
As ever, these things are a pain to download, but considering how much there was, it went pretty well.
I've got a server running on my laptop.
They provide some tools for editing their gazebo virtual environments, but actually using gazebo for editing is awkward.
The agent, brain and environment communicate through python transfer functions.
These are a bit awkward, but I managed to get it workin.
I went through several agents, and have several up on the net here .
They all use a simple modification of their environment, their Husky robot, and Nest. (Andrew Rowley mentioned they use neural voltage instead of spikes, but I think that's an easy fix.)
There's an agent that turns left, then right, then moves forward, an agent that parses a command then does these actions (preset in the transfer function), an agent that uses neural vision to turn toward the blue box, and one that allows you to change a text file to continually enter commands.
That last one is a CABot2. (It doesn't have any real memory.)
They don't currently have support for SpiNNaker. They said it would be slow when it comes back.