
There's two parts to a send away 1) sending them in a straight line in a direction away from you 2) controlling when/where they stop.
I've used working trials and gundog training, and have to say the gundog way makes more sense to me.
Taking the second bit first, the stop is trained to whistle at heel from being a young pup, one slow peeeep and dog sits to heel at your side, then it's a case of building up distance.
The second part is done using retrieves, initially from sending your dog out to retrieve something, to building up trust and sending your dog in long straight lines, using a wall/fence helps to keep them straight for training; the dog trusts that there will be something out there for them to retrieve, and if needs be, you tell them where it is with a hunt command.
For working trials send aways, we used a grid of up to 9 cones, to lift dogs and send them on diagonals, as well as left, right, out and of course, back towards you. That was initially trained using up to three cones in a line and food rewards, which were given to the dogs by people based at the cones without any interaction, until the dog had mastered directions, when it was moved onto a grid of six cones and eventually nine.