Скачать или смотреть Time-Triggered Obstacle Avoidance for Autonomous Humanoid Robots

In real-time computer systems the response time is a critical aspect for time dependent applications. The delay between obstacle detection and the action must be minimised to guarantee the correct obstacle avoidance behaviour to the system. The two mandatory characteristics required to classify the system as real-time environment are response time deadline achievement and system behaviour correctness. The deadlines is defined by time where the result must be available for corrective actions~\cite{Kopetz:TheBook}. The main hard deadline discussed in this article is defined by the obstacle detection cycle duration plus the duration of the corrective action before the needed gap turning or stopping the robot, with no collision. The time-triggered architecture is used in this experiment to follow periodic monitoring events, evaluate the situation and trigger corrective actions. The distance between the robot and the obstacle will be the monitored real time entity triggering the associated corrective actions. In this article, we will discuss about different obstacle avoidance approaches, real time systems and experiment a time-triggered approach to provide consistent obstacle avoidance using a simplified but efficient method in the Nao Aldebaran humanoid robot.