论文部分内容阅读
在动态对抗性环境下竞争的多机器人系统,对于对手的意图进行有效而准确的识别,从而有针对性地提出对策,是获得优势的必要条件。该文针对多机器人系统的对手意图识别提出了基于行为的行为序列估计和行为规则拟合的二次估计方法,引入环境约束、危险评价和行为概率作为判据对多个意图进行排序,避免单一行为结果的推断,实时准确地为决策提供支持。通过机器人足球这一典型的对抗性多机器人系统平台进行了验证,证明该方法在实时性和准确性上均比现有方法有很大提高。
In a multi-robot system that competes in a dynamic confrontational environment, the effective and accurate identification of the opponent’s intention is to propose the countermeasures in a targeted manner, which is the prerequisite for obtaining the advantage. Aiming at the opponent intention recognition of multi-robot system, this paper proposed a second-order estimation method based on behavioral behavioral sequence estimation and behavioral rules fitting. It introduced environmental constraints, risk assessment and behavioral probabilities as criteria to sort multiple intentions to avoid single Extrapolation of behavioral outcomes to support decision-making in real time. It is verified by robot football, a typical confrontational multi-robotic system platform, that this method has a great improvement in real-time and accuracy compared with the existing methods.