Implementation of Longitudinal Control Algorithm for Vehicle Merging

1 I n t r o d u c t i o n Automated highway systems (AHS) offer the potential of significant improvements in traffic flow volume and stability, as well as safety, when compared with current manually-controlled highway driving (Shladover, 1991). Implementation of AHS requires completely automated lateral and longitudinal control of vehicle motions as well as coordination of the maneuvers of different vehicles. Considerable success has already been demonstrated in achieving high-performance automatic lateral (Peng and Tomitzuka, 1994; Tan et al., 1999) and longitudinal (Hedrick, 1998; Lu et al., 2000) control of full-scale road vehicles on test track. The merge maneuver requires a combination of decisions and control actions at both the regulation layer and the coordination layer of the five-layer control hierarchy that Varaiya defined for AHS (Varaiya, 1993). Current thinking about AHS operations favors the addition of vehicles entering at a merge junction on to the back end of a passing platoon of vehicles for reasons of safety and operating simplicity. However, a more general operating condition, which could provide higher efficiency in high-density traffic conditions, would permit entering vehicles to be inserted into the middle of a passing platoon. This more challenging case has been implemented here to show that it is feasible. The simpler case (attaching to the back of a passing platoon) requires only a subset of the capability demonstrated here. Most generally, vehicle merging can be abstracted as the problem of one vehicle from the entry lane merging between two vehicles traveling in a platoon in the main lane. Other situations are all special cases of this general case. Generally, a roadside control computer at the merge junction would perform the coordination layer tasks, but in the experimental implementation described here those functions were performed by the lead vehicle in the main lane. These tasks involve: (a) the determination of vehicle ID of the vehicles in the platoon before and after the merging maneuver (b) the selection of two vehicles in the main lane between which the merging vehicle is to enter (based on the traffic situation in both main lane and merging lane and the road geometry) (c) the coordination between the platoons in the main lane (d) passing relevant information to each vehicle. The regulation layer tasks are fulfilled by each vehicle itself, and always involve the lower-level control of throttle, brake and steering actuation. In some cases, they may also involve generating higher-level control commands based on reference trajectories (although these may be considered as coordination-layer activities in some implementations). Prior research on vehicle merging has addressed the design of the coordination layer protocols (Antoniotti et al., 1999) for simulation. Related research works are referred to (Narendran et al., 1992). However, the problem addressed and the solution obtained here are different. This paper concentrates on the implementation of a newly developed algorithm for general merging of vehicles into a highway (Lu and Hedrick, 2000a). This completely new real-time closed-loop adaptive merging algorithm generates a smooth reference trajectory for the merging vehicle according to the speed of the leader vehicle in the platoon in the main lane. It is not only suitable for two typical road layouts which represent most practical cases but is also applicable to the difficult situation when the speed of the platoon vp(t) in the main lane is changing with respect to time. Part of the main idea here is to introduce the concept of virlual platooning which effectively shifts the time of platoon formation forward prior to the start of real merging. This is the highlight of the algorithm, and is extremely important for real-time implementation and safety. One vehicle merging in between two other vehicles requires that at the time instant of merging Tmerg, the following two conditions are to be satisfied: (i)