The field of swarm robotics studies bio-inspired cooperative control strategies for large groups of relatively simple robots. The robots are limited in their individual capabilities, however, by inducing cooperation amongst them, the limitations can be overcome. Local sensing and interactions within the robotic swarm promote scalable, robust, and flexible behaviours. This thesis focuses on synthesising and analysing minimalist control strategies for swarm robotic systems. Using a computation-free swarming framework, multiple decentralised control strategies are synthesised and analysed. The control strategies enable the robots—equipped with only discrete-valued sensors—to reactively respond to their environment. We present the simplest control solutions to date to four multi-agent problems: finding consensus, gathering on a grid, shepherding, and spatial coverage. The control solutions—obtained by employing an offline evolutionary robotics approach—are tested, either in computer simulation or by physical experiment. They are shown to be—up to a certain extent—scalable, robust against sensor noise, and flexible to the changes in their environment. The investigated gathering problem is proven to be unsolvable using the deterministic framework. The extended framework, using stochastic reactive controllers, is applied to obtain provably correct solutions. Using no run-time memory and only limited sensing make it possible to realise implementations that are arguably free of arithmetic computation. Due to the low computational demands, the control solutions may enable or inspire novel applications, for example, in nanomedicine.