KidVO: a kinodynamically consistent algorithm for online motion planning in dynamic environments

Abstract

Purpose – The purpose of this paper is to propose an efficient method, called kinodynamic velocity obstacle (KidVO), for motion planning of omnimobile robots considering kinematic and dynamic constraints (KDCs). Design/methodology/approach – The suggested method improves generalized velocity obstacle (GVO) approach by a systematic selection of proper time horizon. Selection procedure of the time horizon is based on kinematical and dynamical restrictions of the robot. Toward this aim, an omnimobile robot with a general geometry is taken into account, and the admissible velocity and acceleration cones reflecting KDCs are derived, respectively. To prove the advantages of the suggested planning method, its performance is compared with GVOs, the so-called Hamilton-Jacobi-Bellman equation and the rapidly exploring random tree. Findings – The obtained results of the presented scenarios which contain both computer and real-world experiments for complicated crowded environments indicate the merits of the suggested methodology in terms of its near-optimal behavior, successful obstacle avoidance both in static and dynamic environments and reaching to the goal pose. Originality/value – This paper proposes a novel method for online motion planning of omnimobile robots in dynamic environments while considering the real capabilities of the robot.

A survey on routing problems and robotic systems

“…As far as the reactive approaches are considered, the original VO concept was later extended to handle robots subject to non-holonomic kinematic constraints. 65,104 However, although attainable by the robot, paths planned by on-line navigation strategies may fall short with respect to generating the shortest path.…”

Multi-robot navigation based on velocity obstacle prediction in dynamic crowded environments

“…With respect to solving higher-order kinematic constraints (Arul and Manocha, 2021), buffered Voronoi unit cells were superimposed on RVO velocity cones (V-RVO) to predict the appropriate orientation for each robot by selecting appropriate target points on the cell boundaries. Mutual collision avoidance between robots is realized in car-like robot environments (Alonso-Mora et al , 2012), ORCA for multiple non-holonomic robots (NH-ORCA) (Alonso-Mora et al , 2013; Mahmoodi et al , 2016) and heterogeneous robot environments(Alonso-Mora et al , 2018) by considering non-holonomic kinematic constraints through control inputs of optimal tracking based on optimal interaction collision avoidance, respectively.…”