A Bio-inspired Intelligent Approach to Motion Planning for Mobile Robots

Abstract

Although huge potentials abound for mobile robots in areas like military, traffic control, logistics, and computer graphics, practicable motion planning techniques have yet to be developed. This paper proposes a bio-inspired intelligent approach to motion planning for decentralised mobile objects in dynamic environments. It is inspired by the natural behaviours of creatures which tend to keep a safe distance between one another, and move towards their respective destinations. The proposed approach does not require any central controller, and there is no communication between robots. It instead imitates the characteristics of creatures with local sensing for detection of imminent neighbours and navigation. Each robot is assumed to be driven by a virtual attractive force of its destination and repulsive forces of its imminent neighbours. In particular, it features a module to detect imminent neighbours, reducing computation overheads and eliminating redundant robot movements. Moreover, in comparison with other methods that are mostly based on a simple function for virtual force calculation, the proposed approach adopts a more adaptive two-section function to calculate repulsive forces to improve collision avoidance. A simulator, with a case study of motion planning for free-range automated guided vehicles at a container terminal, is developed to implement and validate the proposed approach.

Keywords:

• bio-inspired
• mobile robots
• motion planning
• dynamic environments