With the advent of autonomous driving, concepts like road trains or platoons are becoming more popular. In these arrangements, vehicles travel at separations of only 5 to 10m between them. These short inter-vehicle distances allow compacting vehicle flows resulting in increased throughput on highways. In addition, there are also fuel/energy savings as the magnitude of aerodynamic resistance acting on vehicles is reduced. These benefits increase when reducing inter-vehicle separations to below 5m. However, it becomes extremely difficult to guarantee safety, especially, when braking in an emergency. The longitudinal and lateral control systems developed so far aim to achieve string stability in the cruise scenario, i.e., to prevent that small variations at the lead magnify towards the trail. Unfortunately, this has no relevance during emergency braking, since control systems incur saturation, i.e., the condition where computed output brake forces exceed those that can be applied by actuators. This is because all vehicles have to apply their maximum brake forces in order to minimize the stopping distance of the platoon and reach a complete standstill. As a result, emergency braking requires special attention and needs to be designed and verified independent of the cruise scenario. Braking in an emergency is mainly characterized by the problem of heterogeneous deceleration capabilities of vehicles, e.g., due to their type and/or loading conditions. As a result, a deceleration rate possible by one vehicle may not be achievable by its immediately leading or following vehicles. Not addressing this heterogeneity leads to inter-vehicle collisions. Moreover, transitions in the road profile increase the complexity of such brake maneuvers. Particularly, when there is a transition from a flat road to a steep downhill, an already saturated brake controller cannot counteract the effect of the downhill slope. Hence, its deceleration magnitude will be reduced, potentially leading to intra-platoon crashes that would otherwise not occur on a flat road. In this work, we first analyze the problem of emergency braking in platoons operating at inter-vehicle separations below 5m and under idealized conditions (i.e., flat road, instantaneous deceleration, etc.). For this case, we propose a cyber-physical approach based on exploiting space buffers that are present in the separations between vehicles, and compare it with straightforward schemes (such as Least Platoon Length and Least Stopping Distance) in terms of achieved aerodynamic benefits, overall platoon length, and stopping distance. We then consider realistic conditions (in particular, changing road profiles as mentioned before) and investigate how to design a brake-by-wire controller present at each vehicle that accounts for this. We further extend our proposed cyber-physical approach by adding cooperative behavior. In particular, if an individual vehicle is unable to track its assigned deceleration, it coordinates with all others to avoid inter-vehicle collisions, for which we propose a vehicle-to-vehicle (V2V) communication strategy. Finally, we present a detailed evaluation of the proposed cyber-physical approach based on high-fidelity vehicle models in Matlab/Simulink. Even though more work is needed towards a real-life implementation, our simulation results demonstrate benefits by the proposed approach and, especially, its feasibility.