Human behavior when navigating a dynamic environment relies on predicting the near future that describes immediate states of moving objects based on their current trajectories. Mobile robots, instead of using a similar approach in many cases, rely only on feedback from the current moment in time, which can result in close encounters with obstacles, leading to trajectory instability due to a strong control signal coming from avoidance algorithms. This paper presents an approach that uses predictive control based on dynamic environment estimation performed by an Unscented Kalman Filter. The predictor analyzes trajectories of nearby objects and assesses the possibility of collision or close encounter. This introduces an additional control signal integrated into the Artificial Potential Fields algorithm, which helps to reduce rapid changes in the control signal, in effect smoothening the robot’s trajectory, as well as reducing the likelihood of dangerous situations happening.
Bonar et al. (Fri,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: