Overtaking opponents with blocking strategies using fuzzy logic

Abstract

In car racing, blocking refers to maneuvers that can prevent, disturb or possibly block an overtaking action by an incoming car. In this paper, we present an advanced overtaking behavior that is able to deal with opponents implementing advanced blocking strategies. The behavior we developed has been integrated in an existing fuzzy-based architecture for driving simulated cars and tested using The Open Car Racing Simulator (TORCS). We compared a driver implementing our overtaking strategy against six of the bots available in the TORCS distribution and simplix, a state-of-the-art bot which won the 2009 TORCS Endurance World Championship. The comparison was carried out against opponents implementing three blocking strategies of increasing difficulty. The results we present show that our strategy can overtake the opponent car in all the considered scenarios. In contrast, all the other bots can complete the overtaking maneuvers in only less than 40% of the cases. Our strategy is slightly more risky than others and may result in limited rear and lateral damage. Other more cautious drivers receive almost no damage, however they can overtake only around 30% of the cases.

Learning overtaking and blocking skills in simulated car racing

“…The overtaking strategy is represented by a set of hand-crafted fuzzy logic rules. Experiments in [14] indicate that, while all the built-in AI-cars fail badly to overtake opponents with blocking strategies, AI-cars equipped with the overtaking strategies in this paper succeed most of the time. The experiments in [14] are only limited to long straight sections of race tracks, leaving some doubts as to whether the fuzzy-logic overtaking strategies are still useful in more realistic races.…”

“…Among the many research papers on TORCS and car racing games, [13] and [14] are two that mainly deal with overtaking. The main idea of [13] is to use Q-learning to acquire overtaking skills.…”

“…The focus of [14] is on overtaking while the opponent has a blocking strategy, which is not covered by [13]. Three blocking strategies are examined: limited blocking, slowly reactive, and fully reactive.…”

“…Overall, both [13] and [14] deal with overtaking in very limited scenarios, and questions remain regarding whether those overtaking skills are applicable in more general situations. Generalization is an important issue in any learning system, as the system learned may be of little use if it only works well in scenarios it has faced during the learning process.…”

An evolutionary tuned driving system for virtual car racing games: The AUTOPIA driver

“…The work presented in Ref. 6 is focused on blocking behaviors and how to deal with them during a race by means of the use of fuzzy logic. In Ref.…”