Architectures for vehicle control, including systems for safety maneuvers and for autonomous driving, need to handle emergency actions. In critical situations, this includes both emergency braking and emergency avoidance. For emergency braking, ABS strives to maximize braking forces, and does so without knowledge about actual friction. For emergency avoidance, recent research has shown analogous results, where algorithms strive to maximize avoidance forces without knowledge about actual friction. The principles of such algorithms are presented, and they perform well in both simulation and experiments. It gives a unified perspective on architectures for emergency braking and avoidance, and it provides a control approach when driving, or handling situations, at the limit of friction.