A hierarchical Model Predictive Control framework for on-road formation control of autonomous vehicles

Abstract

This paper presents an approach for the formation control of autonomous vehicles traversing along a multi-lane road with obstacles and traffic. A major challenge in this problem is a requirement for integrating individual vehicle behaviors such as lane-keeping and collision avoidance with a global formation maintenance behavior. We propose a hierarchical Model Predictive Control (MPC) approach. The desired formation is modeled as a virtual structure evolving curvilinearly along a centerline, and vehicle configurations are expressed as curvilinear relative longitudinal and lateral offsets from the virtual center. At high-level, the trajectory generation of the virtual center is achieved through an MPC framework, which allows various on-road driving constraints to be considered in the optimization. At low-level, a local MPC controller computes the vehicle inputs in order to track the desired trajectory, taking into account more personalized driving constraints. High-fidelity simulations show that the proposed approach drives vehicles to the desired formation while retains some freedom for individual vehicle behaviors.

A Reinforcement Learning Based Approach for Automated Lane Change Maneuvers

“…This is also a clear limitation in the current time-receding optimization methods such as Model Predicted Control (MPC) as the optimization criteria may be too complex to be explicitly formulated for all scenarios, and such methods always involves the predictions of future trajectories. An alternative way is to connect all vehicles via cooperative techniques such as in [6], but in the present paper we deal with a stand-alone system.…”

“…To assess the overall performance, we also need to calculate the total reward that is a cumulative return of immediate rewards over a complete lane changing process. Equally, the total reward can also be viewed as a composition of the three aforementioned individual parts: the total reward from yaw acceleration ,778 , the total reward from yaw rate ?,@8 , and the total reward from lane changing time @AB8 , as shown in (6). = ,778 A J AKL + ?,@8 A J AKL + @AB8 A J AKL (6) In the formulation, we define the rewards with negative values, also called "cost".…”

“…Equally, the total reward can also be viewed as a composition of the three aforementioned individual parts: the total reward from yaw acceleration ,778 , the total reward from yaw rate ?,@8 , and the total reward from lane changing time @AB8 , as shown in (6). = ,778 A J AKL + ?,@8 A J AKL + @AB8 A J AKL (6) In the formulation, we define the rewards with negative values, also called "cost". The idea is that a cost can be considered as a penalty of an action if we want to evaluate the adverse impact of an action, such as unsmooth or inefficient of a lane change maneuver.…”

Cooperative Automated Maneuvering at the 2016 Grand Cooperative Driving Challenge

“…It should be noted that in the context of the GCDC, it was not feasible to apply an architecture that supports formationlike controllers [30], [31] or even a more generic approach based on consensus seeking [32], since this would require all vehicles to implement exactly the same type of vehicle motion controller, whereas in the GCDC, the control approach could be freely chosen by each participant. Nevertheless, the presented architecture does assume all vehicles to be cooperative.…”

Partitioning of the free space-time for on-road navigation of autonomous ground vehicles

“…Even with simpler dynamic models, the nonconvexity of the state-space renders continuous optimization techniques inefficient. For this reason, hierarchical frameworks [5], [6] have been proposed, in which a medium-term (up to a dozen seconds) planner generates a rough trajectory which is then refined by a short-term (sub-second to a 1 few seconds) controller. Mixed-integer programming (MIP) methods are often used in medium-term trajectory planning to encode the discrete decisions arising from multiple maneuver choices [7], [8], generalized as logical constraints in [9].…”