Safe Automated Lane Merging in Mixed Traffic by using Model Predictive Control

Automated vehicles come with the promise that they will improve mobility and transportation systems, making them safer, more comfortable, more efficient and more sustainable. Additionally, for an increasing level of automation (Level 4+) they will release the drivers of their driving tasks and allow them to perform secondary activities during the drive. Bringing these vehicles on the road will require a transition where autonomous vehicles will coexist with manually driven vehicles, so called mixed traffic. As a result, one of the greatest challenges identified in the literature is to plan safe and optimal trajectories for automated vehicles in complex and uncertain situations, with prior unknown movements of manually driven vehicles. Previous research mentions that future road users’ predictions are a prerequisite to achieve safe and high-quality motion planning for autonomous driving. This is a possible claim, yet no research exists that explains to what extend the availability prediction is useful. To quantifiable investigate the claim that other road users’ predictions indeed result in safer and comfortable traffic, in this thesis a safety and comfort analysis of different highway scenarios is performed for different levels of automation. Our simulation-based analysis shows that the availability of other road users’ predictions indeed leads to increased safety in different lane merging scenarios (i.e., merging and traffic jam; short and long ramps). However, comfort improvements with predictions was not witnessed in the data, due to a balance between comfort and safety (i.e., cases that become safe are not necessarily comfortable). To predict future movements of manually driven vehicles, combined data-driven and model-based methods could be used. When multiple predictions are available (e.g., future states as predicted by behavior models (BM) or external predictions (EP) from prediction algorithms), the challenge is to take advantage of all predictions and find an optimal trajectory and control strategy. Thereto, this thesis presents a model predictive control (MPC)-framework that combines EP with MPC’s BM and analyzes the effect of combining multiple predictions. The framework is applied to and analyzed for a two-vehicle, multi-lane merging scenario. It is shown that the framework inherits proactive performance properties due to the EP and can anticipate better on future motions of other road users, which is lacking in the MPC with only BM. The integrated framework is also able to use the BM target vehicle cooperative behavior, which is lacking when only using EP. Therefore, the framework is able to handle a wider variety of scenarios and improve traffic behavior. However, given these predictions with varying accuracy, challenge remains to find an optimal trajectory and a control strategy for the automated Level 4+ vehicle in every road situation. One of these challenging road situations is during lane merging, which is the focus of the present thesis. Lane change is challenging, as multiple tasks have to executed at once, e.g., cognition (detecting empty space), decision making (decide on the merging gap) and driving operations (drive car to gap). When the merging algorithm is not proactively able to make a decision in time and find a gap, the autonomous vehicle can freeze and stop functioning, with all corresponding risks involved. Therefore, we propose a novel mixed integer quadratic program (MIQP) MPC-based solution with recursive feasibility guarantees; robust MPC with tightened constraint set with robust recursive feasibility guarantees; and a contingency MPC-based solution combining Gaussian Process(GP)-MPC with robust MPC, to perform a merging procedure from a double lane into a single lane and continue with (adaptive) cruise control ((A)CC) functionality after the merge in one integrated algorithm. The proposed MPCs balance fast progress along the path with comfort, while obeying a state-dependent safety distance and velocity bounds. Through numerical simulations and, in case of the MIQP-MPC and robust MPC, multi-brand real sized vehicle test-track experiments, we demonstrate the behavior and effectiveness of the proposed MPC merging schemes.