Tactical cooperative planning for autonomous highway driving using Monte-Carlo Tree Search

Abstract

Human drivers use nonverbal communication and anticipation of other drivers' actions to master conflicts occurring in everyday driving situations. Without a high penetration of vehicle-to-vehicle communication an autonomous vehicle has to have the possibility to understand intentions of others and share own intentions with the surrounding traffic participants. This paper proposes a cooperative combinatorial motion planning algorithm without the need for inter vehicle communication based on Monte Carlo Tree Search (MCTS). We motivate why MCTS is particularly suited for the autonomous driving domain. Furthermore, adoptions to the MCTS algorithm are presented as for example simultaneous decisions, the usage of the Intelligent Driver Model as microscopic traffic simulation, and a cooperative cost function. We further show simulation results of merging scenarios in highway-like situations to underline the cooperative nature of the approach.

Planning and Decision-Making for Autonomous Vehicles

“…Since each agent's reward depends not only on its own reward and actions but also on all other agents' actions at the previous stages, the tree grows exponentially with the number of agents. To achieve faster optimization for an optimal (or approximately optimal) solution, other tree search algorithms, such as Monte Carlo tree search (92), may be applied. To reduce computational complexity, Schwarting & Pascheka (93) assumed that the following vehicles' actions are dominated by their predecessors and used this assumption to formulate a recursive conflict-resolution algorithm to achieve only quadratic complexity in the number of agents.…”

From Specifications to Behavior: Maneuver Verification in a Semantic State Space

“…The problem of behavior planning is challenging due to the high-dimensional continuous state space, non-linear motion dynamics, interactions with other agents and their unobservable intentions. This has led to researchers investigating collaborative approaches [6], game-theoretic approaches [7], as well as probabilistic approaches [8]. However, since these new approaches try to incorporate reactive predictions into the decision making process, the complexity is growing and it becomes even harder to fuse them with classic rule-based systems to abide by the traffic regulations and specifications.…”

Decentralized Cooperative Planning for Automated Vehicles with Hierarchical Monte Carlo Tree Search

“…The potential of MCTS for cooperative driving is first presented in [16]. Based on Information-Set MCTS presented in ( [17], [18]) they ensure decoupled decision making and conduct decentralized planning.…”