Real-Time Motion-Planning in Dynamic Environments via Enhanced Velocity Obstacle

Abstract

Freely networked production lines are the essence of the factories of the future. In this context, the mobility of the robotic agents (i.e. mobile manipulators) and logistic units for transferring the goods and the products in a highly dynamic environment shall be considered as a challenging problem. The challenge lies mainly on the recognition of the changes in the environment and fast proper reaction to these changes. The available solutions to the problem are mostly based on the information from embedded local sensoric systems of the mobile units. In this paper, we present a solution based on a global networked sensoric system. We use the depth cameras to monitor the environment of a population of the robotic agents. We apply enhanced method of the Velocity Obstacle to perform a collision free path planning for mobile agents. To save space and computation time, we enhance the collision cones of the obstacle velocities by defining oblique ellipsoidal cones instead of straight circular cones. The aim is to make the process for real-time scenarios as simple, quick and practical as possible. The described methods will be implemented in a dynamic laboratory and the results will be discussed.