I've been stuck with this question for a while. I'm aware of the fact that in circuit switching all the packets use the same path, but does circuit switching need to know the network structure in order to function properly?
Depending on the the specific technology, circuit switching doesn't have "packets." The circuit is a direct point to point link between the endpoints. In the case where data is packetized, there is still a single point to point link. In either case, there is no need for the data to "know" how to get from one end to the other.
Circuit switching requires an initial step - called signaling - which is used to set up the circuit. Signaling does need to understand the topology so the correct devices can be configured to create the point to point circuit. Once the circuit is established, the data flows along the circuit. Signaling is used again at the end to tear down the circuit and make the resources available for another use.
With circuit switching, all packets that belong to an established connection use the same path (if packets are used at all). Of course, there needs to be some kind of routing = finding the path from one connection partner to the other. The methods can vary significantly, depending on the protocol. Routing is often hierarchical, ie. it takes place in central nodes that control the infrastructural, subordinate nodes.
With circuit switching, a connection is only really routed while it is established. Once the connection is established, the transit nodes know which way to forward each packet. This is in stark contrast to packet-switched networks where the transit nodes (bridges and routers) are essentially stateless and forward each packet individually, without using a connection concept.