Pellets (with deuterium and tritium or other fusion gases) are laid down in an outbound track for the spacecraft that will eventually use them, deployed in advance by small spacecraft seeding the runway along the route of flight. Kare thinks in terms of a runway about half a light-day in length. The accelerating spacecraft would gobble up the fusion pellets one at a time, taking about ten days to exit the Solar System, moving along a runway track that stretched from near Earth to beyond the orbit of Pluto.
The vehicle in the shape of a doughnut, or perhaps in more elongated form as a cylinder. The spacecraft would have its own supply of fusion fuel pellets. As the craft accelerates, it drops a pellet into the central ‘hole’ when one of the pellets of the fusion runway is about to be encountered. Nearing the end of the fusion runway, the spacecraft is being driven by fusion explosions at the rate of thirty per second.
The fusion runway relies on impact fusion, with the departing spacecraft first needing to reach speeds high enough (about 200 kilometers per second, by Kare’s reckoning) to ignite the reaction. Once ignition is achieved, the craft continues to accelerate along the runway track. The runway length would have to be adjusted depending on the mission, with robotic probes obviously capable of coping with far higher accelerations than humans. Add a human crew at 1 g of acceleration and a fusion runway might need to stretch out to a tenth of a light year.
String enough fuel pellets along the runway and the spacecraft gets up to ten percent of c. Although Kare built his workshop presentation around a one-ton interstellar probe, he sees the concept as scalable. Larger ships can be accelerated with more pellets.