Do you have SSH access to a remote cluster with precious compute power and want to run a policy in the range of 5 Hz that is resource intensive?
Choose a local port you want to forward to a remote port, e.g. LOCAL_PORT = CLUSTER_PORT = 9999 and connect via ssh:
ssh -L LOCAL_PORT:localhost:CLUSTER_PORT user@cluster
pip install ssh-stream
On the client (your robot @inference)
from ssh_stream import Client
client = Client(port=9999)
client.connect()
while True:
client.send("Hello World")
client.close()On the cluster
from ssh_stream import Server
server = Server(ip='0.0.0.0', port=9999)
server.start()