Reciprocal Velocity Obstacles for real-time multi-agent navigation

Abstract

Abstract-In this paper, we propose a new conceptthe "Reciprocal Velocity Obstacle"-for real-time multi-agent navigation. We consider the case in which each agent navigates independently without explicit communication with other agents. Our formulation is an extension of the Velocity Obstacle concept [3], which was introduced for navigation among (passively) moving obstacles. Our approach takes into account the reactive behavior of the other agents by implicitly assuming that the other agents make a similar collision-avoidance reasoning. We show that this method guarantees safe and oscillationfree motions for each of the agents. We apply our concept to navigation of hundreds of agents in densely populated environments containing both static and moving obstacles, and we show that real-time and scalable performance is achieved in such challenging scenarios.

A Generic Bi-Layer Data-Driven Crowd Behaviors Modeling Approach

“…The other class of the existing works focuses on designing automatic methods to calibrate model parameters [32], [33], [39]. Optimization algorithm such as the Gradientbased Newton method and the Genetic Algorithm are used to calibrate parameters of SFM [2] and the reciprocal velocity obstacles model [33].…”

“…Optimization algorithm such as the Gradientbased Newton method and the Genetic Algorithm are used to calibrate parameters of SFM [2] and the reciprocal velocity obstacles model [33]. For example, Yamaguchi et al [32] designed an automatic method to estimate the setting of personal and social factors in their behavior model with machine learning techniques.…”

“…In this paper, the RVO2, which was proposed by Guy et al in 2008, is adopted, owing to its highly effectiveness in generating realistic collision avoidance behaviors [33], [38]. In the RVO2, for each agent, the observed velocities of other agents are taken into account to select an optimal velocity.…”

On Combining Multi-robot Coverage and Reciprocal Collision Avoidance

“…We build on the aforementioned example of CVT-based Voronoi coverage and combine it with reciprocal collision avoidance in velocity space, using the RVO and ORCA 3 methods [9][10][11][12].…”

“…The vectorsv = v H i − v H j form the set of relative velocities between the robots, τ is the time horizon for a collision to occur and D(p , r) = {q | q − p 2 < r} is the open ball of radius r. The RVO and ORCA methods [9,10] now assume that all the robots make similar attempts in order to avoid collisions. The set of collision-free velocities ORCA τ i| j for a robot i with respect to any other robot j in its neighborhood results from VO τ i| j through an adjustment in velocity by…”