Lateral and longitudinal control of electric bus to follow leader vehicle trajectory

Buses face problems when the capacity of a bus is limited but it should be larger to be able to carry more passengers. The capacity of a bus is already increased to its maximum that is allowed by the infrastructure. The capacity of a bus line could be increased by driving buses more frequently but it would increase costs, that is unwanted. Costs could be reduced by driving buses as platoons consisting of two buses where only the first bus would be operated by a professional driver and the second would be driven autonomously. Autonomous driving requires longitudinal and lateral control of a vehicle. The follower bus should be able to follow the path driven by the leader bus precisely and avoid inter-vehicular collisions while still driving as close together as possible to indicate other traffic that they move as a platoon. Lateral control is usually divided into path following and direct following methods in the literature. Path following methods include obtaining the path of the leader vehicle and following of that path. Path following methods are usually accurate in terms of lateral error but are complex and require a lot of computational capacity. Direct following methods are easy to compute but they do not guarantee precise path following. A simulation model consisting of two identical buses was developed. One longitudinal controller and four lateral control laws were proposed. Longitudinal controller was designed to work also in tight turns which is not usually investigated. Lateral control laws proposed were geometrical in nature and required only input as the relative position of the leader bus. Therefore, they did not require inter-vehicular communication. Longitudinal controller worked well for initialization of the system with inter-vehicular distances from 2 to 10 m. It worked well in acceleration and deceleration tests when both buses were loaded similarly but failed to prevent collisions when follower bus was loaded more heavily than the leader. In lateral controller tests, Pure Pursuit and Modified Pure Pursuit methods were able to follow the leader producing following lateral errors: 0,8 m and 1,1 m (steady-state tests), 0,8 m and 0,7 m (u-turn maneuver) and 0,3 m/0,4 m and 0,4 m/0,5 m (double lane change maneuver, 5 m/s/10 m/s respectively). Spline Pursuit method showed oscillatory behavior and did not follow the leader well. Circular Pursuit method showed also oscillatory behavior and did not follow the leader well. However, it showed better performance than the Spline Pursuit. It remains to be studied whether Pure Pursuit or Modified Pure Pursuit can challenge more sophisticated path following methods.