Starships are usually imagined as gigantic contraptions. This is a guarantee they'll never be undertaken. Instead, the author proposes a "starship" with a weight of 3-8 kg and the payload of 250-750 g.
Propulsion is performed by the following ingenious mechanism. While radioactive decay is symmetric over the solid angle, geometric placement of the fissile material (e.g. Am-241 or Pu-240) can be achieved in such a way that the decay products from a tablet of such material in the direction of the motion of the spaceship will be reabsorbed in the elements of the construction. On the contrary, the decay products in the direction of motion will be emitted predominantly into free space. Reaction force then will accelerate the vehicle.
Fission alone cannot achieve anywhere near a few percent of c needed for the interstellar travel. Yet, the ship will have the sails made of Pd foil saturated with D, T or He3 or a Li6 foil. Fast particles, which are produced by the fission reaction will (with some probability) produce fusion in the asymmetric sails. Again, the sails (author presumes) can be built is such a way that a net recoil from a fusion reaction will be directed along the acceleration path.
The space system thus constructed can be made so cheap that the swarm of these vehicles with AI-divided tasks between them can accomplish investigation of the planetary system (almost) as successfully as a large probe.
The payload (similar to the Coke can) will have microscopic actuators to tighten or slacken the wires to the sails. Because the maximum thrust of the contraption is expected to be on the order of a fraction of a newton, the actuators to direct the ship need not be particularly strong. Of course, acceleration/deceleration of such ship is more or less stochastic.
Prior art: 1) Suggestions of L. Alvarez (one of the fathers of US hydrogen bomb and asteroid hypothesis for the Jurassic collapse) to use fission of transuranium elements emitting into gas jet for space propulsion.
2) Tritium based initiators for nuclear explosions.
3) Stochastic micromachines or nanobots.
Unsolved problem: the communication of the data back to Earth from another planetary system. When the apparatus reaches the nearest stars with planetary systems (in 150-400 years; no reason to be worried of obsolescence--Voyager is now 37 years old), there will be need to transmit the data to the Earth if anyone will still be there to listen with the transmission system weighting on the order of 150-250 g and the available energy on the order of a few watts.